mirror of
https://github.com/introlab/rtabmap.git
synced 2026-09-02 17:40:23 +08:00
2634 lines
86 KiB
C++
2634 lines
86 KiB
C++
/*
|
|
Copyright (c) 2010-2016, Mathieu Labbe - IntRoLab - Universite de Sherbrooke
|
|
All rights reserved.
|
|
|
|
Redistribution and use in source and binary forms, with or without
|
|
modification, are permitted provided that the following conditions are met:
|
|
* Redistributions of source code must retain the above copyright
|
|
notice, this list of conditions and the following disclaimer.
|
|
* Redistributions in binary form must reproduce the above copyright
|
|
notice, this list of conditions and the following disclaimer in the
|
|
documentation and/or other materials provided with the distribution.
|
|
* Neither the name of the Universite de Sherbrooke nor the
|
|
names of its contributors may be used to endorse or promote products
|
|
derived from this software without specific prior written permission.
|
|
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
|
|
ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
|
|
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
|
|
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY
|
|
DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
|
|
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
|
|
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
|
|
ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
|
|
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
*/
|
|
|
|
#include <tango-gl/conversions.h>
|
|
|
|
#include "RTABMapApp.h"
|
|
|
|
#include <rtabmap/core/Rtabmap.h>
|
|
#include <rtabmap/core/util2d.h>
|
|
#include <rtabmap/core/util3d.h>
|
|
#include <rtabmap/core/util3d_transforms.h>
|
|
#include <rtabmap/core/util3d_filtering.h>
|
|
#include <rtabmap/core/util3d_surface.h>
|
|
#include <rtabmap/core/Graph.h>
|
|
#include <rtabmap/utilite/UEventsManager.h>
|
|
#include <rtabmap/utilite/UStl.h>
|
|
#include <rtabmap/utilite/UDirectory.h>
|
|
#include <rtabmap/utilite/UFile.h>
|
|
#include <opencv2/opencv_modules.hpp>
|
|
#include <rtabmap/core/util3d_surface.h>
|
|
#include <rtabmap/utilite/UConversion.h>
|
|
#include <rtabmap/utilite/UTimer.h>
|
|
#include <rtabmap/core/ParamEvent.h>
|
|
#include <rtabmap/core/Compression.h>
|
|
#include <rtabmap/core/Optimizer.h>
|
|
#include <rtabmap/core/VWDictionary.h>
|
|
#include <rtabmap/core/Memory.h>
|
|
#include <rtabmap/core/GainCompensator.h>
|
|
#include <pcl/filters/extract_indices.h>
|
|
#include <pcl/io/ply_io.h>
|
|
#include <pcl/io/obj_io.h>
|
|
#include <pcl/surface/poisson.h>
|
|
#include <pcl/surface/vtk_smoothing/vtk_mesh_quadric_decimation.h>
|
|
|
|
|
|
const int g_exportedMeshId = -100;
|
|
|
|
static JavaVM *jvm;
|
|
static jobject RTABMapActivity = 0;
|
|
|
|
namespace {
|
|
constexpr int kTangoCoreMinimumVersion = 9377;
|
|
} // anonymous namespace.
|
|
|
|
rtabmap::ParametersMap RTABMapApp::getRtabmapParameters()
|
|
{
|
|
rtabmap::ParametersMap parameters;
|
|
|
|
parameters.insert(mappingParameters_.begin(), mappingParameters_.end());
|
|
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kKpMaxFeatures(), std::string("200")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kGFTTQualityLevel(), std::string("0.0001")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kMemImagePreDecimation(), std::string(fullResolution_?"2":"1")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kBRIEFBytes(), std::string("64")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kRtabmapTimeThr(), std::string("800")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kRtabmapPublishLikelihood(), std::string("false")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kRtabmapPublishPdf(), std::string("false")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kRtabmapStartNewMapOnLoopClosure(), uBool2Str(appendMode_)));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kMemBinDataKept(), uBool2Str(!trajectoryMode_)));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kOptimizerIterations(), graphOptimization_?"10":"0"));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kMemIncrementalMemory(), uBool2Str(!localizationMode_)));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kRtabmapMaxRetrieved(), "1"));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kRGBDMaxLocalRetrieved(), "0"));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kKpMaxDepth(), std::string("10"))); // to avoid extracting features in invalid depth (as we compute transformation directly from the words)
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kRGBDOptimizeFromGraphEnd(), std::string("true")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kDbSqlite3InMemory(), std::string("true")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kVisMinInliers(), std::string("25")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kVisEstimationType(), std::string("0"))); // 0=3D-3D 1=PnP
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kRGBDOptimizeMaxError(), std::string("0.1")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kRGBDProximityPathMaxNeighbors(), std::string("0"))); // disable scan matching to merged nodes
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kRGBDProximityBySpace(), std::string("false"))); // just keep loop closure detection
|
|
|
|
if(parameters.find(rtabmap::Parameters::kOptimizerStrategy()) != parameters.end())
|
|
{
|
|
if(parameters.at(rtabmap::Parameters::kOptimizerStrategy()).compare("2") == 0) // GTSAM
|
|
{
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kOptimizerEpsilon(), "0.00001"));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kOptimizerIterations(), graphOptimization_?"10":"0"));
|
|
}
|
|
else if(parameters.at(rtabmap::Parameters::kOptimizerStrategy()).compare("1") == 0) // g2o
|
|
{
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kOptimizerEpsilon(), "0.0"));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kOptimizerIterations(), graphOptimization_?"10":"0"));
|
|
}
|
|
else // TORO
|
|
{
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kOptimizerEpsilon(), "0.00001"));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kOptimizerIterations(), graphOptimization_?"100":"0"));
|
|
}
|
|
}
|
|
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kIcpPointToPlane(), std::string("true")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kMemLaserScanNormalK(), std::string("10")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kIcpIterations(), std::string("10")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kIcpEpsilon(), std::string("0.001")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kIcpMaxRotation(), std::string("0.17"))); // 10 degrees
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kIcpMaxTranslation(), std::string("0.05")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kIcpCorrespondenceRatio(), std::string("0.5")));
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kIcpMaxCorrespondenceDistance(), std::string("0.05")));
|
|
|
|
parameters.insert(*rtabmap::Parameters::getDefaultParameters().find(rtabmap::Parameters::kKpMaxFeatures()));
|
|
parameters.insert(*rtabmap::Parameters::getDefaultParameters().find(rtabmap::Parameters::kMemRehearsalSimilarity()));
|
|
parameters.insert(*rtabmap::Parameters::getDefaultParameters().find(rtabmap::Parameters::kMemMapLabelsAdded()));
|
|
if(dataRecorderMode_)
|
|
{
|
|
uInsert(parameters, rtabmap::ParametersPair(rtabmap::Parameters::kKpMaxFeatures(), std::string("-1")));
|
|
uInsert(parameters, rtabmap::ParametersPair(rtabmap::Parameters::kMemRehearsalSimilarity(), std::string("1.0"))); // deactivate rehearsal
|
|
uInsert(parameters, rtabmap::ParametersPair(rtabmap::Parameters::kMemMapLabelsAdded(), "false")); // don't create map labels
|
|
uInsert(parameters, rtabmap::ParametersPair(rtabmap::Parameters::kMemNotLinkedNodesKept(), std::string("true")));
|
|
}
|
|
|
|
return parameters;
|
|
}
|
|
|
|
RTABMapApp::RTABMapApp() :
|
|
camera_(0),
|
|
rtabmapThread_(0),
|
|
rtabmap_(0),
|
|
logHandler_(0),
|
|
odomCloudShown_(true),
|
|
graphOptimization_(true),
|
|
nodesFiltering_(false),
|
|
localizationMode_(false),
|
|
trajectoryMode_(false),
|
|
autoExposure_(true),
|
|
rawScanSaved_(false),
|
|
smoothing_(true),
|
|
fullResolution_(false),
|
|
appendMode_(true),
|
|
maxCloudDepth_(0.0),
|
|
meshDecimation_(1),
|
|
meshTrianglePix_(1),
|
|
meshAngleToleranceDeg_(15.0),
|
|
minClusterSize_(200),
|
|
maxGainRadius_(0.02f),
|
|
paused_(false),
|
|
dataRecorderMode_(false),
|
|
clearSceneOnNextRender_(false),
|
|
optimizeOpenedDatabase_(true),
|
|
filterPolygonsOnNextRender_(false),
|
|
gainCompensationOnNextRender_(0),
|
|
bilateralFilteringOnNextRender_(false),
|
|
cameraJustInitialized_(false),
|
|
totalPoints_(0),
|
|
totalPolygons_(0),
|
|
lastDrawnCloudsCount_(0),
|
|
renderingTime_(0.0f),
|
|
processMemoryUsedBytes(0),
|
|
processGPUMemoryUsedBytes(0),
|
|
visualizingMesh_(false),
|
|
exportedMeshUpdated_(false),
|
|
exportedMesh_(new pcl::TextureMesh),
|
|
mapToOdom_(rtabmap::Transform::getIdentity())
|
|
|
|
{
|
|
mappingParameters_.insert(rtabmap::ParametersPair(rtabmap::Parameters::kKpDetectorStrategy(), "5")); // GFTT/FREAK
|
|
}
|
|
|
|
RTABMapApp::~RTABMapApp() {
|
|
if(camera_)
|
|
{
|
|
delete camera_;
|
|
}
|
|
if(rtabmapThread_)
|
|
{
|
|
rtabmapThread_->close(false);
|
|
delete rtabmapThread_;
|
|
}
|
|
if(logHandler_)
|
|
{
|
|
delete logHandler_;
|
|
}
|
|
}
|
|
|
|
void RTABMapApp::onCreate(JNIEnv* env, jobject caller_activity)
|
|
{
|
|
env->GetJavaVM(&jvm);
|
|
RTABMapActivity = env->NewGlobalRef(caller_activity);
|
|
|
|
LOGI("RTABMapApp::onCreate()");
|
|
createdMeshes_.clear();
|
|
rawPoses_.clear();
|
|
clearSceneOnNextRender_ = true;
|
|
totalPoints_ = 0;
|
|
totalPolygons_ = 0;
|
|
lastDrawnCloudsCount_ = 0;
|
|
renderingTime_ = 0.0f;
|
|
processMemoryUsedBytes = 0;
|
|
processGPUMemoryUsedBytes = 0;
|
|
|
|
if(camera_)
|
|
{
|
|
delete camera_;
|
|
camera_ = 0;
|
|
}
|
|
if(rtabmapThread_)
|
|
{
|
|
rtabmapThread_->close(false);
|
|
delete rtabmapThread_;
|
|
rtabmapThread_ = 0;
|
|
rtabmap_ = 0;
|
|
}
|
|
|
|
if(logHandler_ == 0)
|
|
{
|
|
logHandler_ = new LogHandler();
|
|
}
|
|
|
|
this->registerToEventsManager();
|
|
|
|
camera_ = new rtabmap::CameraTango(fullResolution_?1:2, autoExposure_, rawScanSaved_, smoothing_);
|
|
}
|
|
|
|
void RTABMapApp::setScreenRotation(int displayRotation, int cameraRotation)
|
|
{
|
|
TangoSupportRotation rotation = tango_gl::util::GetAndroidRotationFromColorCameraToDisplay(
|
|
displayRotation, cameraRotation);
|
|
LOGI("Set orientation: display=%d camera=%d -> %d", displayRotation, cameraRotation, (int)rotation);
|
|
main_scene_.setScreenRotation(rotation);
|
|
camera_->setScreenRotation(rotation);
|
|
}
|
|
|
|
int RTABMapApp::openDatabase(const std::string & databasePath, bool databaseInMemory, bool optimize)
|
|
{
|
|
this->unregisterFromEventsManager(); // to ignore published init events when closing rtabmap
|
|
status_.first = rtabmap::RtabmapEventInit::kInitializing;
|
|
rtabmapMutex_.lock();
|
|
if(rtabmapThread_)
|
|
{
|
|
rtabmapThread_->close(false);
|
|
delete rtabmapThread_;
|
|
rtabmapThread_ = 0;
|
|
rtabmap_ = 0;
|
|
}
|
|
|
|
//Rtabmap
|
|
mapToOdom_.setIdentity();
|
|
rtabmap_ = new rtabmap::Rtabmap();
|
|
rtabmap::ParametersMap parameters = getRtabmapParameters();
|
|
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kDbSqlite3InMemory(), uBool2Str(databaseInMemory)));
|
|
rtabmap_->init(parameters, databasePath);
|
|
rtabmapThread_ = new rtabmap::RtabmapThread(rtabmap_);
|
|
if(parameters.find(rtabmap::Parameters::kRtabmapDetectionRate()) != parameters.end())
|
|
{
|
|
rtabmapThread_->setDetectorRate(uStr2Float(parameters.at(rtabmap::Parameters::kRtabmapDetectionRate())));
|
|
}
|
|
|
|
// Generate all meshes
|
|
std::map<int, rtabmap::Signature> signatures;
|
|
std::map<int, rtabmap::Transform> poses;
|
|
std::multimap<int, rtabmap::Link> links;
|
|
rtabmap_->get3DMap(
|
|
signatures,
|
|
poses,
|
|
links,
|
|
true,
|
|
true);
|
|
|
|
int status = 0;
|
|
if(signatures.size() && poses.empty())
|
|
{
|
|
LOGE("Failed to optimize the graph!");
|
|
status = -1;
|
|
}
|
|
|
|
optimizeOpenedDatabase_ = optimize;
|
|
clearSceneOnNextRender_ = true;
|
|
rtabmap::Statistics stats;
|
|
stats.setSignatures(signatures);
|
|
stats.addStatistic(rtabmap::Statistics::kMemoryWorking_memory_size(), (float)rtabmap_->getWMSize());
|
|
stats.addStatistic(rtabmap::Statistics::kKeypointDictionary_size(), (float)rtabmap_->getMemory()->getVWDictionary()->getVisualWords().size());
|
|
stats.addStatistic(rtabmap::Statistics::kMemoryDatabase_memory_used(), (float)rtabmap_->getMemory()->getDatabaseMemoryUsed());
|
|
stats.setPoses(poses);
|
|
stats.setConstraints(links);
|
|
|
|
LOGI("Open: add rtabmap event with loaded data");
|
|
rtabmapEvents_.push_back(stats);
|
|
|
|
// Start threads
|
|
LOGI("Start rtabmap thread");
|
|
this->registerToEventsManager();
|
|
rtabmapThread_->registerToEventsManager();
|
|
rtabmapThread_->start();
|
|
|
|
status_.first = rtabmap::RtabmapEventInit::kInitialized;
|
|
status_.second = "";
|
|
rtabmapMutex_.unlock();
|
|
|
|
return status;
|
|
}
|
|
|
|
bool RTABMapApp::onTangoServiceConnected(JNIEnv* env, jobject iBinder)
|
|
{
|
|
LOGW("onTangoServiceConnected()");
|
|
if(camera_)
|
|
{
|
|
camera_->join(true);
|
|
|
|
if (TangoService_setBinder(env, iBinder) != TANGO_SUCCESS) {
|
|
UERROR("TangoHandler::ConnectTango, TangoService_setBinder error");
|
|
return false;
|
|
}
|
|
|
|
if(camera_->init())
|
|
{
|
|
LOGI("Start camera thread");
|
|
if(!paused_)
|
|
{
|
|
camera_->start();
|
|
}
|
|
cameraJustInitialized_ = true;
|
|
return true;
|
|
}
|
|
UERROR("Failed camera initialization!");
|
|
}
|
|
return false;
|
|
}
|
|
|
|
void RTABMapApp::onPause()
|
|
{
|
|
LOGW("onPause()");
|
|
if(camera_)
|
|
{
|
|
camera_->join(true);
|
|
camera_->close();
|
|
}
|
|
}
|
|
|
|
|
|
void RTABMapApp::TangoResetMotionTracking() {
|
|
TangoService_resetMotionTracking();
|
|
}
|
|
|
|
// OpenGL thread
|
|
void RTABMapApp::InitializeGLContent()
|
|
{
|
|
UINFO("");
|
|
main_scene_.InitGLContent();
|
|
}
|
|
|
|
// OpenGL thread
|
|
void RTABMapApp::SetViewPort(int width, int height)
|
|
{
|
|
UINFO("");
|
|
main_scene_.SetupViewPort(width, height);
|
|
}
|
|
|
|
class PostRenderEvent : public UEvent
|
|
{
|
|
public:
|
|
PostRenderEvent(const rtabmap::Statistics & stats) :
|
|
stats_(stats)
|
|
{
|
|
|
|
}
|
|
virtual std::string getClassName() const {return "PostRenderEvent";}
|
|
const rtabmap::Statistics & getStats() const {return stats_;}
|
|
private:
|
|
rtabmap::Statistics stats_;
|
|
};
|
|
|
|
// OpenGL thread
|
|
bool RTABMapApp::smoothMesh(int id, Mesh & mesh)
|
|
{
|
|
UTimer t;
|
|
// reconstruct depth image
|
|
UASSERT(mesh.indices.get() && mesh.indices->size());
|
|
cv::Mat depth = cv::Mat::zeros(mesh.cloud->height, mesh.cloud->width, CV_32FC1);
|
|
rtabmap::Transform localTransformInv = mesh.cameraModel.localTransform().inverse();
|
|
for(unsigned int i=0; i<mesh.indices->size(); ++i)
|
|
{
|
|
int index = mesh.indices->at(i);
|
|
// FastBilateralFilter works in camera frame
|
|
if(mesh.cloud->at(index).x > 0)
|
|
{
|
|
pcl::PointXYZRGB pt = rtabmap::util3d::transformPoint(mesh.cloud->at(index), localTransformInv);
|
|
depth.at<float>(index) = pt.z;
|
|
}
|
|
}
|
|
|
|
depth = rtabmap::util2d::fastBilateralFiltering(depth, 2.0f, 0.075f);
|
|
LOGI("smoothMesh() Bilateral filtering of %d, time=%fs", id, t.ticks());
|
|
|
|
if(!depth.empty() && mesh.indices->size())
|
|
{
|
|
pcl::IndicesPtr newIndices(new std::vector<int>(mesh.indices->size()));
|
|
int oi = 0;
|
|
for(unsigned int i=0; i<mesh.indices->size(); ++i)
|
|
{
|
|
int index = mesh.indices->at(i);
|
|
|
|
pcl::PointXYZRGB & pt = mesh.cloud->at(index);
|
|
pcl::PointXYZRGB newPt = rtabmap::util3d::transformPoint(mesh.cloud->at(index), localTransformInv);
|
|
if(depth.at<float>(index) > 0)
|
|
{
|
|
newPt.z = depth.at<float>(index);
|
|
newPt = rtabmap::util3d::transformPoint(newPt, mesh.cameraModel.localTransform());
|
|
newIndices->at(oi++) = index;
|
|
}
|
|
else
|
|
{
|
|
newPt.x = newPt.y = newPt.z = std::numeric_limits<float>::quiet_NaN();
|
|
}
|
|
pt.x = newPt.x;
|
|
pt.y = newPt.y;
|
|
pt.z = newPt.z;
|
|
}
|
|
newIndices->resize(oi);
|
|
mesh.indices = newIndices;
|
|
|
|
//reconstruct the mesh with smoothed surfaces
|
|
std::vector<pcl::Vertices> polygons;
|
|
if(main_scene_.isMeshRendering())
|
|
{
|
|
polygons = rtabmap::util3d::organizedFastMesh(mesh.cloud, meshAngleToleranceDeg_*M_PI/180.0, false, meshTrianglePix_);
|
|
}
|
|
LOGI("smoothMesh() Reconstructing the mesh of %d, time=%fs", id, t.ticks());
|
|
mesh.polygons = polygons;
|
|
}
|
|
else
|
|
{
|
|
UERROR("smoothMesh() Failed to smooth surface %d", id);
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
|
|
// OpenGL thread
|
|
int RTABMapApp::Render()
|
|
{
|
|
try
|
|
{
|
|
UASSERT(camera_!=0 && rtabmap_!=0);
|
|
|
|
UTimer fpsTime;
|
|
boost::mutex::scoped_lock lock(renderingMutex_);
|
|
|
|
if(clearSceneOnNextRender_)
|
|
{
|
|
visualizingMesh_ = false;
|
|
}
|
|
|
|
bool notifyCameraStarted = false;
|
|
|
|
// process only pose events in vsualization mode
|
|
rtabmap::Transform pose;
|
|
{
|
|
boost::mutex::scoped_lock lock(poseMutex_);
|
|
if(poseEvents_.size())
|
|
{
|
|
pose = poseEvents_.back();
|
|
poseEvents_.clear();
|
|
}
|
|
}
|
|
rtabmap::Transform mapOdom = rtabmap::Transform::getIdentity();
|
|
if(!pose.isNull())
|
|
{
|
|
// update camera pose?
|
|
if(graphOptimization_ && !visualizingMesh_ && !mapToOdom_.isIdentity())
|
|
{
|
|
mapOdom = mapToOdom_;
|
|
main_scene_.SetCameraPose(opengl_world_T_rtabmap_world*mapOdom*rtabmap_world_T_tango_world*pose);
|
|
}
|
|
else
|
|
{
|
|
main_scene_.SetCameraPose(opengl_world_T_tango_world*pose);
|
|
}
|
|
if(!camera_->isRunning() && cameraJustInitialized_)
|
|
{
|
|
notifyCameraStarted = true;
|
|
cameraJustInitialized_ = false;
|
|
}
|
|
}
|
|
|
|
rtabmap::OdometryEvent odomEvent;
|
|
{
|
|
boost::mutex::scoped_lock lock(odomMutex_);
|
|
if(odomEvents_.size())
|
|
{
|
|
LOGI("Process odom events");
|
|
odomEvent = odomEvents_.back();
|
|
odomEvents_.clear();
|
|
if(cameraJustInitialized_)
|
|
{
|
|
notifyCameraStarted = true;
|
|
cameraJustInitialized_ = false;
|
|
}
|
|
}
|
|
}
|
|
|
|
if(visualizingMesh_)
|
|
{
|
|
if(exportedMeshUpdated_)
|
|
{
|
|
main_scene_.clear();
|
|
exportedMeshUpdated_ = false;
|
|
}
|
|
if(!main_scene_.hasCloud(g_exportedMeshId))
|
|
{
|
|
if(exportedMesh_->tex_polygons.size() && exportedMesh_->tex_polygons[0].size())
|
|
{
|
|
Mesh mesh;
|
|
mesh.gain = 1.0f;
|
|
mesh.cloud.reset(new pcl::PointCloud<pcl::PointXYZRGB>);
|
|
mesh.normals.reset(new pcl::PointCloud<pcl::Normal>);
|
|
pcl::fromPCLPointCloud2(exportedMesh_->cloud, *mesh.cloud);
|
|
pcl::fromPCLPointCloud2(exportedMesh_->cloud, *mesh.normals);
|
|
mesh.polygons = exportedMesh_->tex_polygons[0];
|
|
cv::Mat texture;
|
|
if(exportedMesh_->tex_coordinates.size())
|
|
{
|
|
mesh.texCoords = exportedMesh_->tex_coordinates[0];
|
|
texture = exportedTexture_;
|
|
}
|
|
|
|
main_scene_.addMesh(g_exportedMeshId, mesh, texture, opengl_world_T_rtabmap_world);
|
|
}
|
|
else
|
|
{
|
|
pcl::IndicesPtr indices(new std::vector<int>); // null
|
|
pcl::PointCloud<pcl::PointXYZRGB>::Ptr cloud(new pcl::PointCloud<pcl::PointXYZRGB>);
|
|
pcl::fromPCLPointCloud2(exportedMesh_->cloud, *cloud);
|
|
main_scene_.addCloud(g_exportedMeshId, cloud, indices, opengl_world_T_rtabmap_world);
|
|
}
|
|
}
|
|
|
|
//backup state
|
|
bool isMeshRendering = main_scene_.isMeshRendering();
|
|
bool isTextureRendering = main_scene_.isMeshTexturing();
|
|
bool isFrustumCulling = main_scene_.isFrustumCulling();
|
|
|
|
main_scene_.setMeshRendering(main_scene_.hasMesh(g_exportedMeshId), main_scene_.hasTexture(g_exportedMeshId));
|
|
main_scene_.setFrustumCulling(false);
|
|
|
|
main_scene_.Render();
|
|
|
|
// revert state
|
|
main_scene_.setMeshRendering(isMeshRendering, isTextureRendering);
|
|
main_scene_.setFrustumCulling(isFrustumCulling);
|
|
|
|
if(renderingTime_ < fpsTime.elapsed())
|
|
{
|
|
renderingTime_ = fpsTime.elapsed();
|
|
}
|
|
|
|
return notifyCameraStarted?1:0;
|
|
}
|
|
else
|
|
{
|
|
if(main_scene_.hasCloud(g_exportedMeshId))
|
|
{
|
|
main_scene_.clear();
|
|
exportedMesh_.reset(new pcl::TextureMesh);
|
|
exportedTexture_ = cv::Mat();
|
|
}
|
|
|
|
bool notifyDataLoaded = false;
|
|
// should be before clearSceneOnNextRender_ in case openDatabase is called
|
|
std::list<rtabmap::Statistics> rtabmapEvents;
|
|
{
|
|
boost::mutex::scoped_lock lock(rtabmapMutex_);
|
|
rtabmapEvents = rtabmapEvents_;
|
|
rtabmapEvents_.clear();
|
|
|
|
boost::mutex::scoped_lock lockMesh(meshesMutex_);
|
|
if(!clearSceneOnNextRender_ && rtabmapEvents.size() && createdMeshes_.size())
|
|
{
|
|
if(rtabmapEvents.front().refImageId()>0 && rtabmapEvents.front().refImageId() < createdMeshes_.rbegin()->first)
|
|
{
|
|
LOGI("Detected new database! new=%d old=%d", rtabmapEvents.front().refImageId(), createdMeshes_.rbegin()->first);
|
|
clearSceneOnNextRender_ = true;
|
|
}
|
|
}
|
|
}
|
|
|
|
if(clearSceneOnNextRender_)
|
|
{
|
|
boost::mutex::scoped_lock lock(meshesMutex_);
|
|
|
|
odomMutex_.lock();
|
|
odomEvents_.clear();
|
|
odomMutex_.unlock();
|
|
|
|
poseMutex_.lock();
|
|
poseEvents_.clear();
|
|
poseMutex_.unlock();
|
|
|
|
main_scene_.clear();
|
|
clearSceneOnNextRender_ = false;
|
|
createdMeshes_.clear();
|
|
rawPoses_.clear();
|
|
totalPoints_ = 0;
|
|
totalPolygons_ = 0;
|
|
lastDrawnCloudsCount_ = 0;
|
|
renderingTime_ = 0.0f;
|
|
processMemoryUsedBytes = 0;
|
|
processGPUMemoryUsedBytes = 0;
|
|
}
|
|
|
|
// Did we lose OpenGL context? If so, recreate the context;
|
|
std::set<int> added = main_scene_.getAddedClouds();
|
|
added.erase(-1);
|
|
{
|
|
boost::mutex::scoped_lock lock(meshesMutex_);
|
|
if(added.size() != createdMeshes_.size())
|
|
{
|
|
processGPUMemoryUsedBytes = 0;
|
|
for(std::map<int, Mesh>::iterator iter=createdMeshes_.begin(); iter!=createdMeshes_.end(); ++iter)
|
|
{
|
|
if(!main_scene_.hasCloud(iter->first))
|
|
{
|
|
if(main_scene_.isMeshRendering() && iter->second.polygons.size() == 0)
|
|
{
|
|
iter->second.polygons = rtabmap::util3d::organizedFastMesh(iter->second.cloud, meshAngleToleranceDeg_*M_PI/180.0, false, meshTrianglePix_);
|
|
}
|
|
cv::Mat texture;
|
|
if(main_scene_.isMeshTexturing())
|
|
{
|
|
cv::Mat textureRaw;
|
|
textureRaw = rtabmap::uncompressImage(rtabmap_->getMemory()->getImageCompressed(iter->first));
|
|
if(!textureRaw.empty())
|
|
{
|
|
cv::Size reducedSize(textureRaw.cols/(textureRaw.cols>1000?4:2), textureRaw.rows/(textureRaw.cols>1000?4:2));
|
|
LOGD("resize image from %dx%d to %dx%d", textureRaw.cols, textureRaw.rows, reducedSize.width, reducedSize.height);
|
|
cv::resize(textureRaw, texture, reducedSize, 0, 0, CV_INTER_AREA);
|
|
}
|
|
}
|
|
main_scene_.addMesh(iter->first, iter->second, texture, opengl_world_T_rtabmap_world*iter->second.pose);
|
|
main_scene_.setCloudVisible(iter->first, iter->second.visible);
|
|
|
|
long estimateGPUMem = 0;
|
|
estimateGPUMem += iter->second.cloud->size()*16; // 3*float + 1 float rgb
|
|
estimateGPUMem += iter->second.indices->size()*4; // int
|
|
estimateGPUMem += iter->second.polygons.size()*4*3; // 3 indices per polygon
|
|
|
|
processGPUMemoryUsedBytes += estimateGPUMem + (texture.empty()?0:iter->second.polygons.size()*3*8+texture.total());
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
if(rtabmapEvents.size())
|
|
{
|
|
LOGI("Process rtabmap events");
|
|
|
|
// update buffered signatures
|
|
std::map<int, rtabmap::SensorData> bufferedSensorData;
|
|
if(!trajectoryMode_ && !dataRecorderMode_)
|
|
{
|
|
for(std::list<rtabmap::Statistics>::iterator iter=rtabmapEvents.begin(); iter!=rtabmapEvents.end(); ++iter)
|
|
{
|
|
// Don't create mesh for the last node added if rehearsal happened or if discarded (small movement)
|
|
int smallMovement = (int)uValue(iter->data(), rtabmap::Statistics::kMemorySmall_movement(), 0.0f);
|
|
int rehearsalMerged = (int)uValue(iter->data(), rtabmap::Statistics::kMemoryRehearsal_merged(), 0.0f);
|
|
if(smallMovement == 0 && rehearsalMerged == 0)
|
|
{
|
|
for(std::map<int, rtabmap::Signature>::const_iterator jter=iter->getSignatures().begin(); jter!=iter->getSignatures().end(); ++jter)
|
|
{
|
|
bool dataDetected = false;
|
|
if(!jter->second.sensorData().imageRaw().empty() &&
|
|
!jter->second.sensorData().depthRaw().empty())
|
|
{
|
|
if(!localizationMode_)
|
|
{
|
|
uInsert(bufferedSensorData, std::make_pair(jter->first, jter->second.sensorData()));
|
|
uInsert(rawPoses_, std::make_pair(jter->first, jter->second.getPose()));
|
|
dataDetected = true;
|
|
}
|
|
}
|
|
else if(totalPoints_ == 0 &&
|
|
!jter->second.sensorData().imageCompressed().empty() &&
|
|
!jter->second.sensorData().depthOrRightCompressed().empty())
|
|
{
|
|
uInsert(bufferedSensorData, std::make_pair(jter->first, jter->second.sensorData()));
|
|
uInsert(rawPoses_, std::make_pair(jter->first, jter->second.getPose()));
|
|
if(notifyDataLoaded == false)
|
|
{
|
|
LOGI("Detecting that we are loading a database");
|
|
}
|
|
notifyDataLoaded = true;
|
|
dataDetected = true;
|
|
}
|
|
if(dataDetected)
|
|
{
|
|
processMemoryUsedBytes += jter->second.sensorData().imageCompressed().total();
|
|
processMemoryUsedBytes += jter->second.sensorData().depthOrRightCompressed().total();
|
|
processMemoryUsedBytes += jter->second.sensorData().laserScanCompressed().total();
|
|
processMemoryUsedBytes += jter->second.getWords().size()*4*8;
|
|
processMemoryUsedBytes += jter->second.getWords3().size()*4*4;
|
|
if(!jter->second.getWordsDescriptors().empty())
|
|
{
|
|
processMemoryUsedBytes += jter->second.getWordsDescriptors().size()*(4+jter->second.getWordsDescriptors().begin()->second.total());
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
int loopClosure = (int)uValue(iter->data(), rtabmap::Statistics::kLoopAccepted_hypothesis_id(), 0.0f);
|
|
int rejected = (int)uValue(iter->data(), rtabmap::Statistics::kLoopRejectedHypothesis(), 0.0f);
|
|
if(!paused_ && loopClosure>0)
|
|
{
|
|
main_scene_.setBackgroundColor(0, 0.7f, 0); // green
|
|
}
|
|
else if(!paused_ && rejected>0)
|
|
{
|
|
main_scene_.setBackgroundColor(0, 0.1f, 0); // dark green
|
|
}
|
|
else if(!paused_ && rehearsalMerged>0)
|
|
{
|
|
main_scene_.setBackgroundColor(0, 0, 0.2f); // blue
|
|
}
|
|
else
|
|
{
|
|
main_scene_.setBackgroundColor(0, 0, 0);
|
|
}
|
|
}
|
|
}
|
|
|
|
std::map<int, rtabmap::Transform> poses = rtabmapEvents.back().poses();
|
|
if(!rtabmapEvents.back().mapCorrection().isNull())
|
|
{
|
|
mapToOdom_ = rtabmapEvents.back().mapCorrection();
|
|
}
|
|
|
|
// Transform pose in OpenGL world
|
|
for(std::map<int, rtabmap::Transform>::iterator iter=poses.begin(); iter!=poses.end(); ++iter)
|
|
{
|
|
if(!graphOptimization_)
|
|
{
|
|
std::map<int, rtabmap::Transform>::iterator jter = rawPoses_.find(iter->first);
|
|
if(jter != rawPoses_.end())
|
|
{
|
|
iter->second = opengl_world_T_rtabmap_world*jter->second;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
iter->second = opengl_world_T_rtabmap_world*iter->second;
|
|
}
|
|
}
|
|
|
|
const std::multimap<int, rtabmap::Link> & links = rtabmapEvents.back().constraints();
|
|
if(poses.size())
|
|
{
|
|
//update graph
|
|
main_scene_.updateGraph(poses, links);
|
|
|
|
// update clouds
|
|
boost::mutex::scoped_lock lock(meshesMutex_);
|
|
std::set<std::string> strIds;
|
|
for(std::map<int, rtabmap::Transform>::iterator iter=poses.begin(); iter!=poses.end(); ++iter)
|
|
{
|
|
int id = iter->first;
|
|
if(!iter->second.isNull())
|
|
{
|
|
if(main_scene_.hasCloud(id))
|
|
{
|
|
//just update pose
|
|
main_scene_.setCloudPose(id, iter->second);
|
|
main_scene_.setCloudVisible(id, true);
|
|
std::map<int, Mesh>::iterator meshIter = createdMeshes_.find(id);
|
|
UASSERT(meshIter!=createdMeshes_.end());
|
|
meshIter->second.pose = opengl_world_T_rtabmap_world.inverse()*iter->second;
|
|
meshIter->second.visible = true;
|
|
}
|
|
else if(uContains(bufferedSensorData, id))
|
|
{
|
|
rtabmap::SensorData data = bufferedSensorData.at(id);
|
|
|
|
cv::Mat tmpA, depth;
|
|
data.uncompressData(&tmpA, &depth);
|
|
|
|
if(!data.imageRaw().empty() && !data.depthRaw().empty())
|
|
{
|
|
// Voxelize and filter depending on the previous cloud?
|
|
pcl::PointCloud<pcl::PointXYZRGB>::Ptr cloud;
|
|
pcl::IndicesPtr indices(new std::vector<int>);
|
|
LOGI("Creating node cloud %d (depth=%dx%d rgb=%dx%d)", id, data.depthRaw().cols, data.depthRaw().rows, data.imageRaw().cols, data.imageRaw().rows);
|
|
cloud = rtabmap::util3d::cloudRGBFromSensorData(data, meshDecimation_, maxCloudDepth_, 0, indices.get());
|
|
|
|
if(cloud->size() && indices->size())
|
|
{
|
|
UTimer time;
|
|
std::vector<pcl::Vertices> polygons;
|
|
if(main_scene_.isMeshRendering())
|
|
{
|
|
polygons = rtabmap::util3d::organizedFastMesh(cloud, meshAngleToleranceDeg_*M_PI/180.0, false, meshTrianglePix_);
|
|
LOGI("Creating mesh, %d polygons (%fs)", (int)polygons.size(), time.ticks());
|
|
}
|
|
|
|
if((main_scene_.isMeshRendering() && polygons.size()) || !main_scene_.isMeshRendering())
|
|
{
|
|
totalPolygons_ += polygons.size();
|
|
|
|
std::pair<std::map<int, Mesh>::iterator, bool> inserted = createdMeshes_.insert(std::make_pair(id, Mesh()));
|
|
UASSERT(inserted.second);
|
|
inserted.first->second.cloud = cloud;
|
|
inserted.first->second.indices = indices;
|
|
inserted.first->second.polygons = polygons;
|
|
inserted.first->second.pose = opengl_world_T_rtabmap_world.inverse()*iter->second;
|
|
inserted.first->second.visible = true;
|
|
inserted.first->second.cameraModel = data.cameraModels()[0];
|
|
inserted.first->second.gain = 1.0f;
|
|
cv::Mat texture;
|
|
if(main_scene_.isMeshTexturing())
|
|
{
|
|
cv::Size reducedSize(data.imageRaw().cols/(data.imageRaw().cols>1000?4:2), data.imageRaw().rows/(data.imageRaw().cols>1000?4:2));
|
|
LOGD("resize image from %dx%d to %dx%d", data.imageRaw().cols, data.imageRaw().rows, reducedSize.width, reducedSize.height);
|
|
cv::resize(data.imageRaw(), texture, reducedSize, 0, 0, CV_INTER_AREA);
|
|
}
|
|
main_scene_.addMesh(id, inserted.first->second, texture, iter->second);
|
|
|
|
long estimateCPUMem = 0;
|
|
estimateCPUMem += inserted.first->second.cloud->size()*16; // 3*float + 1 float rgb
|
|
estimateCPUMem += inserted.first->second.indices->size()*4; // int
|
|
estimateCPUMem += inserted.first->second.polygons.size()*4*3; // 3 indices per polygon
|
|
|
|
processMemoryUsedBytes += estimateCPUMem;
|
|
processGPUMemoryUsedBytes += estimateCPUMem + (texture.empty()?0:inserted.first->second.polygons.size()*3*8+texture.total());
|
|
}
|
|
else
|
|
{
|
|
UERROR("No mesh could be created for node %d", id);
|
|
}
|
|
}
|
|
totalPoints_+=indices->size();
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
//filter poses?
|
|
if(poses.size() > 2)
|
|
{
|
|
if(nodesFiltering_)
|
|
{
|
|
for(std::multimap<int, rtabmap::Link>::const_iterator iter=links.begin(); iter!=links.end(); ++iter)
|
|
{
|
|
if(iter->second.type() != rtabmap::Link::kNeighbor)
|
|
{
|
|
int oldId = iter->second.to()>iter->second.from()?iter->second.from():iter->second.to();
|
|
poses.erase(oldId);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
if(poses.size())
|
|
{
|
|
//update cloud visibility
|
|
boost::mutex::scoped_lock lock(meshesMutex_);
|
|
std::set<int> addedClouds = main_scene_.getAddedClouds();
|
|
for(std::set<int>::const_iterator iter=addedClouds.begin();
|
|
iter!=addedClouds.end();
|
|
++iter)
|
|
{
|
|
if(*iter > 0 && poses.find(*iter) == poses.end())
|
|
{
|
|
main_scene_.setCloudVisible(*iter, false);
|
|
std::map<int, Mesh>::iterator meshIter = createdMeshes_.find(*iter);
|
|
UASSERT(meshIter!=createdMeshes_.end());
|
|
meshIter->second.visible = false;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
else
|
|
{
|
|
main_scene_.setCloudVisible(-1, odomCloudShown_ && !trajectoryMode_ && !paused_);
|
|
|
|
//just process the last one
|
|
if(!odomEvent.pose().isNull())
|
|
{
|
|
if(odomCloudShown_ && !trajectoryMode_)
|
|
{
|
|
if(!odomEvent.data().imageRaw().empty() && !odomEvent.data().depthRaw().empty())
|
|
{
|
|
pcl::PointCloud<pcl::PointXYZRGB>::Ptr cloud;
|
|
pcl::IndicesPtr indices(new std::vector<int>);
|
|
cloud = rtabmap::util3d::cloudRGBFromSensorData(odomEvent.data(), meshDecimation_, maxCloudDepth_, 0.0f, indices.get());
|
|
if(cloud->size() && indices->size())
|
|
{
|
|
LOGI("Created odom cloud (rgb=%dx%d depth=%dx%d cloud=%dx%d)",
|
|
odomEvent.data().imageRaw().cols, odomEvent.data().imageRaw().rows,
|
|
odomEvent.data().depthRaw().cols, odomEvent.data().depthRaw().rows,
|
|
(int)cloud->width, (int)cloud->height);
|
|
main_scene_.addCloud(-1, cloud, indices, opengl_world_T_rtabmap_world*mapOdom*odomEvent.pose());
|
|
main_scene_.setCloudVisible(-1, true);
|
|
}
|
|
else
|
|
{
|
|
UERROR("Generated cloud is empty!");
|
|
}
|
|
}
|
|
else
|
|
{
|
|
UERROR("Odom data images are empty!");
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
if((notifyDataLoaded&&optimizeOpenedDatabase_) || gainCompensationOnNextRender_>0)
|
|
{
|
|
UTimer tGainCompensation;
|
|
LOGI("Gain compensation...");
|
|
boost::mutex::scoped_lock lock(meshesMutex_);
|
|
|
|
std::map<int, pcl::PointCloud<pcl::PointXYZRGB>::Ptr > clouds;
|
|
std::map<int, pcl::IndicesPtr> indices;
|
|
for(std::map<int, Mesh>::iterator iter = createdMeshes_.begin(); iter!=createdMeshes_.end(); ++iter)
|
|
{
|
|
clouds.insert(std::make_pair(iter->first, iter->second.cloud));
|
|
indices.insert(std::make_pair(iter->first, iter->second.indices));
|
|
}
|
|
std::map<int, rtabmap::Transform> poses;
|
|
std::multimap<int, rtabmap::Link> links;
|
|
rtabmap_->getGraph(poses, links, true, true);
|
|
if(gainCompensationOnNextRender_ == 2)
|
|
{
|
|
// full compensation
|
|
links.clear();
|
|
for(std::map<int, pcl::PointCloud<pcl::PointXYZRGB>::Ptr>::const_iterator iter=clouds.begin(); iter!=clouds.end(); ++iter)
|
|
{
|
|
int from = iter->first;
|
|
std::map<int, pcl::PointCloud<pcl::PointXYZRGB>::Ptr>::const_iterator jter = iter;
|
|
++jter;
|
|
for(;jter!=clouds.end(); ++jter)
|
|
{
|
|
int to = jter->first;
|
|
links.insert(std::make_pair(from, rtabmap::Link(from, to, rtabmap::Link::kUserClosure, poses.at(from).inverse()*poses.at(to))));
|
|
}
|
|
}
|
|
}
|
|
|
|
UASSERT(maxGainRadius_>0.0f);
|
|
rtabmap::GainCompensator compensator(maxGainRadius_);
|
|
if(clouds.size() > 1 && links.size())
|
|
{
|
|
compensator.feed(clouds, indices, links);
|
|
LOGI("Gain compensation... compute gain: links=%d, time=%fs", (int)links.size(), tGainCompensation.ticks());
|
|
}
|
|
|
|
for(std::map<int, Mesh>::iterator iter = createdMeshes_.begin(); iter!=createdMeshes_.end(); ++iter)
|
|
{
|
|
if(!iter->second.cloud->empty())
|
|
{
|
|
if(clouds.size() > 1 && links.size())
|
|
{
|
|
iter->second.gain = compensator.getGain(iter->first);
|
|
LOGI("%d mesh has gain %f", iter->first, iter->second.gain);
|
|
}
|
|
}
|
|
|
|
main_scene_.updateGain(iter->first, iter->second.gain);
|
|
}
|
|
LOGI("Gain compensation... applying gain: meshes=%d, time=%fs", (int)createdMeshes_.size(), tGainCompensation.ticks());
|
|
|
|
gainCompensationOnNextRender_ = 0;
|
|
notifyDataLoaded = true;
|
|
}
|
|
|
|
if(bilateralFilteringOnNextRender_)
|
|
{
|
|
LOGI("Bilateral filtering...");
|
|
bilateralFilteringOnNextRender_ = false;
|
|
boost::mutex::scoped_lock lock(meshesMutex_);
|
|
for(std::map<int, Mesh>::iterator iter = createdMeshes_.begin(); iter!=createdMeshes_.end(); ++iter)
|
|
{
|
|
if(iter->second.cloud->size() && iter->second.indices->size())
|
|
{
|
|
if(smoothMesh(iter->first, iter->second))
|
|
{
|
|
main_scene_.updateMesh(iter->first, iter->second, cv::Mat());
|
|
}
|
|
}
|
|
}
|
|
notifyDataLoaded = true;
|
|
}
|
|
|
|
if(filterPolygonsOnNextRender_ && minClusterSize_>0)
|
|
{
|
|
LOGI("Polygon filtering...");
|
|
filterPolygonsOnNextRender_ = false;
|
|
boost::mutex::scoped_lock lock(meshesMutex_);
|
|
for(std::map<int, Mesh>::iterator iter = createdMeshes_.begin(); iter!=createdMeshes_.end(); ++iter)
|
|
{
|
|
if(iter->second.polygons.size())
|
|
{
|
|
// filter polygons
|
|
std::vector<std::set<int> > neighbors;
|
|
std::vector<std::set<int> > vertexToPolygons;
|
|
rtabmap::util3d::createPolygonIndexes(
|
|
iter->second.polygons,
|
|
iter->second.cloud->size(),
|
|
neighbors,
|
|
vertexToPolygons);
|
|
std::list<std::list<int> > clusters = rtabmap::util3d::clusterPolygons(
|
|
neighbors,
|
|
minClusterSize_);
|
|
std::vector<pcl::Vertices> filteredPolygons(iter->second.polygons.size());
|
|
int oi=0;
|
|
for(std::list<std::list<int> >::iterator jter=clusters.begin(); jter!=clusters.end(); ++jter)
|
|
{
|
|
for(std::list<int>::iterator kter=jter->begin(); kter!=jter->end(); ++kter)
|
|
{
|
|
filteredPolygons[oi++] = iter->second.polygons.at(*kter);
|
|
}
|
|
}
|
|
filteredPolygons.resize(oi);
|
|
iter->second.polygons = filteredPolygons;
|
|
main_scene_.updateCloudPolygons(iter->first, iter->second.polygons);
|
|
}
|
|
}
|
|
notifyDataLoaded = true;
|
|
}
|
|
|
|
lastDrawnCloudsCount_ = main_scene_.Render();
|
|
if(renderingTime_ < fpsTime.elapsed())
|
|
{
|
|
renderingTime_ = fpsTime.elapsed();
|
|
}
|
|
|
|
if(rtabmapEvents.size())
|
|
{
|
|
// send statistics to GUI
|
|
LOGI("Posting PostRenderEvent!");
|
|
UEventsManager::post(new PostRenderEvent(rtabmapEvents.back()));
|
|
}
|
|
|
|
return notifyDataLoaded||notifyCameraStarted?1:0;
|
|
}
|
|
}
|
|
catch(const std::exception & e)
|
|
{
|
|
UERROR("Exception! msg=\"%s\"", e.what());
|
|
return -1;
|
|
}
|
|
}
|
|
|
|
void RTABMapApp::SetCameraType(
|
|
tango_gl::GestureCamera::CameraType camera_type) {
|
|
main_scene_.SetCameraType(camera_type);
|
|
}
|
|
|
|
void RTABMapApp::OnTouchEvent(int touch_count,
|
|
tango_gl::GestureCamera::TouchEvent event,
|
|
float x0, float y0, float x1, float y1) {
|
|
main_scene_.OnTouchEvent(touch_count, event, x0, y0, x1, y1);
|
|
}
|
|
|
|
void RTABMapApp::setPausedMapping(bool paused)
|
|
{
|
|
{
|
|
boost::mutex::scoped_lock lock(renderingMutex_);
|
|
visualizingMesh_ = false;
|
|
main_scene_.setBackgroundColor(0, 0, 0);
|
|
}
|
|
paused_ = paused;
|
|
if(camera_)
|
|
{
|
|
if(paused_)
|
|
{
|
|
LOGW("Pause!");
|
|
camera_->kill();
|
|
}
|
|
else
|
|
{
|
|
LOGW("Resume!");
|
|
UEventsManager::post(new rtabmap::RtabmapEventCmd(rtabmap::RtabmapEventCmd::kCmdTriggerNewMap));
|
|
camera_->start();
|
|
}
|
|
}
|
|
}
|
|
void RTABMapApp::setMapCloudShown(bool shown)
|
|
{
|
|
main_scene_.setMapRendering(shown);
|
|
}
|
|
void RTABMapApp::setOdomCloudShown(bool shown)
|
|
{
|
|
odomCloudShown_ = shown;
|
|
main_scene_.setTraceVisible(shown);
|
|
}
|
|
void RTABMapApp::setMeshRendering(bool enabled, bool withTexture)
|
|
{
|
|
main_scene_.setMeshRendering(enabled, withTexture);
|
|
}
|
|
void RTABMapApp::setPointSize(float value)
|
|
{
|
|
main_scene_.setPointSize(value);
|
|
}
|
|
void RTABMapApp::setLighting(bool enabled)
|
|
{
|
|
main_scene_.setLighting(enabled);
|
|
}
|
|
void RTABMapApp::setBackfaceCulling(bool enabled)
|
|
{
|
|
main_scene_.setBackfaceCulling(enabled);
|
|
}
|
|
|
|
void RTABMapApp::setLocalizationMode(bool enabled)
|
|
{
|
|
localizationMode_ = enabled;
|
|
this->post(new rtabmap::ParamEvent(rtabmap::Parameters::kMemIncrementalMemory(), uBool2Str(!localizationMode_)));
|
|
}
|
|
void RTABMapApp::setTrajectoryMode(bool enabled)
|
|
{
|
|
trajectoryMode_ = enabled;
|
|
this->post(new rtabmap::ParamEvent(rtabmap::Parameters::kMemBinDataKept(), uBool2Str(!trajectoryMode_)));
|
|
}
|
|
|
|
void RTABMapApp::setGraphOptimization(bool enabled)
|
|
{
|
|
graphOptimization_ = enabled;
|
|
UASSERT(camera_ != 0 && rtabmap_!=0 && rtabmap_->getMemory()!=0);
|
|
if(!camera_->isRunning() && rtabmap_->getMemory()->getLastWorkingSignature()!=0)
|
|
{
|
|
std::map<int, rtabmap::Transform> poses;
|
|
std::multimap<int, rtabmap::Link> links;
|
|
rtabmap_->getGraph(poses, links, true, true);
|
|
if(poses.size())
|
|
{
|
|
boost::mutex::scoped_lock lock(rtabmapMutex_);
|
|
rtabmap::Statistics stats = rtabmap_->getStatistics();
|
|
stats.setPoses(poses);
|
|
stats.setConstraints(links);
|
|
|
|
LOGI("Send rtabmap event to update graph...");
|
|
rtabmapEvents_.push_back(stats);
|
|
|
|
rtabmap_->setOptimizedPoses(poses);
|
|
}
|
|
}
|
|
}
|
|
void RTABMapApp::setNodesFiltering(bool enabled)
|
|
{
|
|
nodesFiltering_ = enabled;
|
|
setGraphOptimization(graphOptimization_); // this will resend the graph if paused
|
|
}
|
|
void RTABMapApp::setGraphVisible(bool visible)
|
|
{
|
|
main_scene_.setGraphVisible(visible);
|
|
main_scene_.setTraceVisible(visible);
|
|
}
|
|
void RTABMapApp::setGridVisible(bool visible)
|
|
{
|
|
main_scene_.setGridVisible(visible);
|
|
}
|
|
|
|
void RTABMapApp::setAutoExposure(bool enabled)
|
|
{
|
|
if(autoExposure_ != enabled)
|
|
{
|
|
autoExposure_ = enabled;
|
|
if(camera_)
|
|
{
|
|
camera_->setAutoExposure(autoExposure_);
|
|
}
|
|
}
|
|
}
|
|
|
|
void RTABMapApp::setRawScanSaved(bool enabled)
|
|
{
|
|
if(rawScanSaved_ != enabled)
|
|
{
|
|
rawScanSaved_ = enabled;
|
|
if(camera_)
|
|
{
|
|
camera_->setRawScanPublished(rawScanSaved_);
|
|
}
|
|
}
|
|
}
|
|
|
|
void RTABMapApp::setFullResolution(bool enabled)
|
|
{
|
|
if(fullResolution_ != enabled)
|
|
{
|
|
fullResolution_ = enabled;
|
|
if(camera_)
|
|
{
|
|
camera_->setDecimation(fullResolution_?1:2);
|
|
}
|
|
rtabmap::ParametersMap parameters;
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kMemImagePreDecimation(), std::string(fullResolution_?"2":"1")));
|
|
this->post(new rtabmap::ParamEvent(parameters));
|
|
}
|
|
}
|
|
|
|
void RTABMapApp::setSmoothing(bool enabled)
|
|
{
|
|
if(smoothing_ != enabled)
|
|
{
|
|
smoothing_ = enabled;
|
|
if(camera_)
|
|
{
|
|
camera_->setSmoothing(smoothing_);
|
|
}
|
|
}
|
|
}
|
|
|
|
void RTABMapApp::setAppendMode(bool enabled)
|
|
{
|
|
if(appendMode_ != enabled)
|
|
{
|
|
appendMode_ = enabled;
|
|
rtabmap::ParametersMap parameters;
|
|
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kRtabmapStartNewMapOnLoopClosure(), uBool2Str(appendMode_)));
|
|
this->post(new rtabmap::ParamEvent(parameters));
|
|
}
|
|
}
|
|
|
|
void RTABMapApp::setDataRecorderMode(bool enabled)
|
|
{
|
|
if(dataRecorderMode_ != enabled)
|
|
{
|
|
dataRecorderMode_ = enabled; // parameters will be set when resuming (we assume we are paused)
|
|
}
|
|
}
|
|
|
|
void RTABMapApp::setMaxCloudDepth(float value)
|
|
{
|
|
maxCloudDepth_ = value;
|
|
}
|
|
|
|
void RTABMapApp::setMeshDecimation(int value)
|
|
{
|
|
LOGW("Set mesh decimation to level %d", value);
|
|
meshDecimation_ = 1;
|
|
if(camera_)
|
|
{
|
|
// Google Tango Tablet 160x90
|
|
// Phab2Pro 240x135
|
|
int width = camera_->getCameraModel().imageWidth()/8;
|
|
int height = camera_->getCameraModel().imageHeight()/8;
|
|
if(value == 3) // high
|
|
{
|
|
if(width % 10 == 0 && height % 10 == 0)
|
|
{
|
|
meshDecimation_ = 10;
|
|
}
|
|
else if(width % 15 == 0 && height % 15 == 0)
|
|
{
|
|
meshDecimation_ = 15;
|
|
}
|
|
else
|
|
{
|
|
UERROR("Could not set decimation to high (size=%dx%d)", width, height);
|
|
}
|
|
}
|
|
else if(value == 2) // medium
|
|
{
|
|
if(width % 5 == 0 && height % 5 == 0)
|
|
{
|
|
meshDecimation_ = 5;
|
|
}
|
|
else
|
|
{
|
|
UERROR("Could not set decimation to medium (size=%dx%d)", width, height);
|
|
}
|
|
}
|
|
else if(value == 1) // low
|
|
{
|
|
if(width % 3 == 0 && width % 3 == 0)
|
|
{
|
|
meshDecimation_ = 3;
|
|
}
|
|
else if(width % 2 == 0 && width % 2 == 0)
|
|
{
|
|
meshDecimation_ = 2;
|
|
}
|
|
else
|
|
{
|
|
UERROR("Could not set decimation to low (size=%dx%d)", width, height);
|
|
}
|
|
}
|
|
}
|
|
UINFO("Set decimation to %d", meshDecimation_);
|
|
}
|
|
|
|
void RTABMapApp::setMeshAngleTolerance(float value)
|
|
{
|
|
meshAngleToleranceDeg_ = value;
|
|
}
|
|
|
|
void RTABMapApp::setMeshTriangleSize(int value)
|
|
{
|
|
meshTrianglePix_ = value;
|
|
}
|
|
|
|
void RTABMapApp::setMinClusterSize(int value)
|
|
{
|
|
minClusterSize_ = value;
|
|
}
|
|
|
|
void RTABMapApp::setMaxGainRadius(float value)
|
|
{
|
|
maxGainRadius_ = value;
|
|
}
|
|
|
|
int RTABMapApp::setMappingParameter(const std::string & key, const std::string & value)
|
|
{
|
|
std::string compatibleKey = key;
|
|
|
|
// Backward compatibility
|
|
std::map<std::string, std::pair<bool, std::string> >::const_iterator iter=rtabmap::Parameters::getRemovedParameters().find(key);
|
|
if(iter != rtabmap::Parameters::getRemovedParameters().end())
|
|
{
|
|
if(iter->second.first)
|
|
{
|
|
// can be migrated
|
|
compatibleKey = iter->second.second;
|
|
LOGW("Parameter name changed: \"%s\" -> \"%s\". Please update the code accordingly. Value \"%s\" is still set to the new parameter name.",
|
|
iter->first.c_str(), iter->second.second.c_str(), value.c_str());
|
|
}
|
|
else
|
|
{
|
|
if(iter->second.second.empty())
|
|
{
|
|
UERROR("Parameter \"%s\" doesn't exist anymore!",
|
|
iter->first.c_str());
|
|
}
|
|
else
|
|
{
|
|
UERROR("Parameter \"%s\" doesn't exist anymore! You may look at this similar parameter: \"%s\"",
|
|
iter->first.c_str(), iter->second.second.c_str());
|
|
}
|
|
}
|
|
}
|
|
|
|
if(rtabmap::Parameters::getDefaultParameters().find(compatibleKey) != rtabmap::Parameters::getDefaultParameters().end())
|
|
{
|
|
LOGI(uFormat("Setting param \"%s\" to \"%s\"", compatibleKey.c_str(), value.c_str()).c_str());
|
|
if(compatibleKey.compare(rtabmap::Parameters::kKpDetectorStrategy()) == 0 &&
|
|
mappingParameters_.at(rtabmap::Parameters::kKpDetectorStrategy()).compare(value) != 0)
|
|
{
|
|
// Changing feature type should reset mapping!
|
|
resetMapping();
|
|
}
|
|
uInsert(mappingParameters_, rtabmap::ParametersPair(compatibleKey, value));
|
|
UEventsManager::post(new rtabmap::ParamEvent(mappingParameters_));
|
|
return 0;
|
|
}
|
|
else
|
|
{
|
|
UERROR(uFormat("Key \"%s\" doesn't exist!", compatibleKey.c_str()).c_str());
|
|
return -1;
|
|
}
|
|
}
|
|
|
|
void RTABMapApp::resetMapping()
|
|
{
|
|
LOGW("Reset!");
|
|
status_.first = rtabmap::RtabmapEventInit::kInitializing;
|
|
status_.second = "";
|
|
|
|
mapToOdom_.setIdentity();
|
|
clearSceneOnNextRender_ = true;
|
|
|
|
UEventsManager::post(new rtabmap::RtabmapEventCmd(rtabmap::RtabmapEventCmd::kCmdResetMemory));
|
|
}
|
|
|
|
void RTABMapApp::save(const std::string & databasePath)
|
|
{
|
|
rtabmapThread_->join(true);
|
|
|
|
// save mapping parameters in the database
|
|
|
|
bool appendModeBackup = appendMode_;
|
|
if(appendMode_)
|
|
{
|
|
appendMode_ = false;
|
|
}
|
|
|
|
bool dataRecorderModeBackup = dataRecorderMode_;
|
|
if(dataRecorderMode_)
|
|
{
|
|
dataRecorderMode_ = false;
|
|
}
|
|
|
|
if(appendModeBackup || dataRecorderModeBackup)
|
|
{
|
|
rtabmap::ParametersMap parameters = getRtabmapParameters();
|
|
rtabmap_->parseParameters(parameters);
|
|
appendMode_ = appendModeBackup;
|
|
dataRecorderMode_ = dataRecorderModeBackup;
|
|
}
|
|
|
|
rtabmap_->close(true, databasePath);
|
|
rtabmap_->init(getRtabmapParameters(), dataRecorderMode_?"":databasePath);
|
|
if(dataRecorderMode_)
|
|
{
|
|
clearSceneOnNextRender_ = true;
|
|
}
|
|
rtabmapThread_->start();
|
|
|
|
}
|
|
|
|
cv::Mat RTABMapApp::mergeTextures(pcl::TextureMesh & mesh, int textureSize) const
|
|
{
|
|
UASSERT(textureSize> 0);
|
|
LOGD("textureSize = %d materials=%d", textureSize, mesh.tex_materials.size());
|
|
cv::Mat globalTexture;
|
|
if(mesh.tex_materials.size() >= 1)
|
|
{
|
|
std::vector<int> textures(mesh.tex_materials.size(), -1);
|
|
cv::Size imageSize;
|
|
int imageType=CV_8UC3;
|
|
for(unsigned int i=0; i<mesh.tex_materials.size(); ++i)
|
|
{
|
|
if(!mesh.tex_materials[i].tex_file.empty() &&
|
|
mesh.tex_polygons[i].size() &&
|
|
uIsInteger(mesh.tex_materials[i].tex_file, false))
|
|
{
|
|
int textureId = uStr2Int(mesh.tex_materials[i].tex_file);
|
|
textures[i] = textureId;
|
|
|
|
if(imageSize.height == 0)
|
|
{
|
|
rtabmap::SensorData data = rtabmap_->getMemory()->getNodeData(textureId);
|
|
|
|
UASSERT(!data.imageCompressed().empty() &&
|
|
data.cameraModels().size()==1 &&
|
|
data.cameraModels()[0].imageHeight()>0);
|
|
imageSize = data.cameraModels()[0].imageSize();
|
|
}
|
|
}
|
|
else
|
|
{
|
|
textures[i] = -1;
|
|
}
|
|
}
|
|
if(textures.size() && imageSize.height>0 && imageSize.width>0)
|
|
{
|
|
float scale = 0.0f;
|
|
std::vector<bool> materialsKept;
|
|
rtabmap::util3d::concatenateTextureMaterials(mesh, imageSize, textureSize, scale, &materialsKept);
|
|
LOGD("scale=%f materials=%d", scale, (int)mesh.tex_materials.size());
|
|
if(scale && mesh.tex_materials.size()==1)
|
|
{
|
|
int cols = float(textureSize)/(scale*imageSize.width);
|
|
|
|
globalTexture = cv::Mat(textureSize, textureSize, imageType, cv::Scalar::all(255));
|
|
|
|
// make a blank texture
|
|
cv::Mat emptyImage(int(imageSize.height*scale), int(imageSize.width*scale), imageType, cv::Scalar::all(255));
|
|
int oi=0;
|
|
for(int i=0; i<(int)textures.size(); ++i)
|
|
{
|
|
if(materialsKept.at(i))
|
|
{
|
|
int u = oi%cols * emptyImage.cols;
|
|
int v = oi/cols * emptyImage.rows;
|
|
UASSERT(u < textureSize-emptyImage.cols);
|
|
UASSERT(v < textureSize-emptyImage.rows);
|
|
if(textures[i]>=0)
|
|
{
|
|
rtabmap::SensorData data = rtabmap_->getMemory()->getNodeData(textures[i]);
|
|
UASSERT_MSG(!data.imageCompressed().empty(), uFormat("id=%d", textures[i]).c_str());
|
|
cv::Mat image;
|
|
data.uncompressDataConst(&image, 0);
|
|
UASSERT(!image.empty());
|
|
cv::Mat resizedImage;
|
|
cv::resize(image, resizedImage, emptyImage.size(), 0.0f, 0.0f, cv::INTER_AREA);
|
|
if(createdMeshes_.find(textures[i]) != createdMeshes_.end() && createdMeshes_.at(textures[i]).gain != 1.0f)
|
|
{
|
|
cv::multiply(resizedImage, createdMeshes_.at(textures[i]).gain, resizedImage);
|
|
}
|
|
UASSERT(resizedImage.type() == globalTexture.type());
|
|
resizedImage.copyTo(globalTexture(cv::Rect(u, v, resizedImage.cols, resizedImage.rows)));
|
|
}
|
|
else
|
|
{
|
|
emptyImage.copyTo(globalTexture(cv::Rect(u, v, emptyImage.cols, emptyImage.rows)));
|
|
}
|
|
++oi;
|
|
}
|
|
}
|
|
}
|
|
else
|
|
{
|
|
UERROR("Failed merging textures");
|
|
}
|
|
}
|
|
else if(textures.size() == 0)
|
|
{
|
|
UERROR("No textures kept!");
|
|
}
|
|
else
|
|
{
|
|
UERROR("No image size set!");
|
|
}
|
|
}
|
|
return globalTexture;
|
|
}
|
|
|
|
bool RTABMapApp::exportMesh(
|
|
const std::string & filePath,
|
|
float cloudVoxelSize,
|
|
bool regenerateCloud,
|
|
bool meshing,
|
|
int textureSize,
|
|
int normalK,
|
|
float maxTextureDistance,
|
|
bool optimized,
|
|
float optimizedVoxelSize,
|
|
int optimizedDepth,
|
|
float optimizedDecimationFactor,
|
|
float optimizedColorRadius,
|
|
bool optimizedCleanWhitePolygons,
|
|
bool optimizedColorWhitePolygons, // not yet used
|
|
bool blockRendering)
|
|
{
|
|
// make sure createdMeshes_ is not modified while exporting! We don't
|
|
// lock the meshesMutex_ because we want to continue rendering.
|
|
|
|
if(blockRendering)
|
|
{
|
|
renderingMutex_.lock();
|
|
main_scene_.clear();
|
|
}
|
|
|
|
bool success = false;
|
|
|
|
try
|
|
{
|
|
std::map<int, rtabmap::Transform> poses;
|
|
std::multimap<int, rtabmap::Link> links;
|
|
rtabmap_->getGraph(poses, links, true, true);
|
|
|
|
//Assemble the meshes
|
|
if(meshing) // Mesh or Texture Mesh
|
|
{
|
|
pcl::PolygonMesh::Ptr polygonMesh(new pcl::PolygonMesh);
|
|
pcl::TextureMesh::Ptr textureMesh(new pcl::TextureMesh);
|
|
cv::Mat globalTexture;
|
|
int totalPolygons = 0;
|
|
{
|
|
if(optimized)
|
|
{
|
|
std::map<int, rtabmap::Transform> cameraPoses;
|
|
std::map<int, rtabmap::CameraModel> cameraModels;
|
|
|
|
UTimer timer;
|
|
LOGI("Assemble clouds (%d)...", (int)poses.size());
|
|
int cloudCount=0;
|
|
pcl::PointCloud<pcl::PointXYZRGBNormal>::Ptr mergedClouds(new pcl::PointCloud<pcl::PointXYZRGBNormal>);
|
|
for(std::map<int, rtabmap::Transform>::iterator iter=poses.begin();
|
|
iter!= poses.end();
|
|
++iter)
|
|
{
|
|
std::map<int, Mesh>::iterator jter = createdMeshes_.find(iter->first);
|
|
pcl::PointCloud<pcl::PointXYZRGB>::Ptr cloud;
|
|
pcl::IndicesPtr indices(new std::vector<int>);
|
|
rtabmap::CameraModel model;
|
|
float gain = 1.0f;
|
|
if(jter != createdMeshes_.end())
|
|
{
|
|
cloud = jter->second.cloud;
|
|
indices = jter->second.indices;
|
|
model = jter->second.cameraModel;
|
|
gain = jter->second.gain;
|
|
}
|
|
else
|
|
{
|
|
rtabmap::SensorData data = rtabmap_->getMemory()->getNodeData(iter->first, true);
|
|
if(!data.imageRaw().empty() && !data.depthRaw().empty() && data.cameraModels().size() == 1)
|
|
{
|
|
cloud = rtabmap::util3d::cloudRGBFromSensorData(data, meshDecimation_, maxCloudDepth_, 0, indices.get());
|
|
model = data.cameraModels()[0];
|
|
}
|
|
}
|
|
if(cloud->size() && indices->size() && model.isValidForProjection())
|
|
{
|
|
pcl::PointCloud<pcl::PointXYZRGB>::Ptr transformedCloud(new pcl::PointCloud<pcl::PointXYZRGB>);
|
|
if(optimizedVoxelSize > 0.0f)
|
|
{
|
|
transformedCloud = rtabmap::util3d::voxelize(cloud, indices, optimizedVoxelSize);
|
|
transformedCloud = rtabmap::util3d::transformPointCloud(transformedCloud, iter->second);
|
|
}
|
|
else
|
|
{
|
|
// it looks like that using only transformPointCloud with indices
|
|
// flushes the colors, so we should extract points before... maybe a too old PCL version
|
|
pcl::copyPointCloud(*cloud, *indices, *transformedCloud);
|
|
transformedCloud = rtabmap::util3d::transformPointCloud(transformedCloud, iter->second);
|
|
}
|
|
|
|
Eigen::Vector3f viewpoint( iter->second.x(), iter->second.y(), iter->second.z());
|
|
pcl::PointCloud<pcl::Normal>::Ptr normals = rtabmap::util3d::computeNormals(transformedCloud, normalK, viewpoint);
|
|
|
|
pcl::PointCloud<pcl::PointXYZRGBNormal>::Ptr cloudWithNormals(new pcl::PointCloud<pcl::PointXYZRGBNormal>);
|
|
pcl::concatenateFields(*transformedCloud, *normals, *cloudWithNormals);
|
|
|
|
if(textureSize == 0 && gain != 1.0f)
|
|
{
|
|
for(unsigned int i=0; i<cloudWithNormals->size(); ++i)
|
|
{
|
|
pcl::PointXYZRGBNormal & pt = cloudWithNormals->at(i);
|
|
pt.r = uchar(std::max(0.0, std::min(255.0, double(pt.r) * gain)));
|
|
pt.g = uchar(std::max(0.0, std::min(255.0, double(pt.g) * gain)));
|
|
pt.b = uchar(std::max(0.0, std::min(255.0, double(pt.b) * gain)));
|
|
}
|
|
}
|
|
|
|
|
|
if(mergedClouds->size() == 0)
|
|
{
|
|
*mergedClouds = *cloudWithNormals;
|
|
}
|
|
else
|
|
{
|
|
*mergedClouds += *cloudWithNormals;
|
|
}
|
|
|
|
cameraPoses.insert(std::make_pair(iter->first, iter->second));
|
|
cameraModels.insert(std::make_pair(iter->first, model));
|
|
|
|
LOGI("Assembled %d points (%d/%d total=%d)", (int)cloudWithNormals->size(), ++cloudCount, (int)poses.size(), (int)mergedClouds->size());
|
|
}
|
|
else
|
|
{
|
|
UERROR("Cloud %d not found or empty", iter->first);
|
|
}
|
|
}
|
|
LOGI("Assembled clouds (%d)... done! %fs (total points=%d)", (int)cameraPoses.size(), timer.ticks(), (int)mergedClouds->size());
|
|
|
|
if(mergedClouds->size())
|
|
{
|
|
// Mesh reconstruction
|
|
LOGI("Mesh reconstruction...");
|
|
pcl::PolygonMesh::Ptr mesh(new pcl::PolygonMesh);
|
|
pcl::Poisson<pcl::PointXYZRGBNormal> poisson;
|
|
poisson.setDepth(optimizedDepth);
|
|
poisson.setInputCloud(mergedClouds);
|
|
poisson.reconstruct(*mesh);
|
|
LOGI("Mesh reconstruction... done! %fs (%d polygons)", timer.ticks(), mesh->polygons.size());
|
|
|
|
if(mesh->polygons.size())
|
|
{
|
|
if(textureSize > 0 && optimizedDecimationFactor > 0.0f)
|
|
{
|
|
#ifndef DISABLE_VTK
|
|
unsigned int count = mesh->polygons.size();
|
|
LOGI("Mesh decimation (factor=%f) from %d polygons...", optimizedDecimationFactor, (int)count);
|
|
|
|
pcl::PolygonMesh::Ptr output(new pcl::PolygonMesh);
|
|
pcl::MeshQuadricDecimationVTK mqd;
|
|
mqd.setTargetReductionFactor(optimizedDecimationFactor);
|
|
mqd.setInputMesh(mesh);
|
|
mqd.process (*output);
|
|
mesh = output;
|
|
|
|
//mesh = rtabmap::util3d::meshDecimation(mesh, decimationFactor);
|
|
// use direct instantiation above to this fix some linker errors on android like:
|
|
// pcl::MeshQuadricDecimationVTK::performProcessing(pcl::PolygonMesh&): error: undefined reference to 'vtkQuadricDecimation::New()'
|
|
// pcl::VTKUtils::mesh2vtk(pcl::PolygonMesh const&, vtkSmartPointer<vtkPolyData>&): error: undefined reference to 'vtkFloatArray::New()'
|
|
|
|
LOGI("Mesh decimated (factor=%f) from %d to %d polygons (%fs)", optimizedDecimationFactor, count, (int)mesh->polygons.size(), timer.ticks());
|
|
if(count < mesh->polygons.size())
|
|
{
|
|
UWARN("Decimated mesh has more polygons than before!");
|
|
}
|
|
#else
|
|
UWARN("RTAB-Map is not built with PCL-VTK module so mesh decimation cannot be used!");
|
|
#endif
|
|
}
|
|
|
|
if(textureSize == 0)
|
|
{
|
|
// colored polygon mesh
|
|
if(optimizedColorRadius >= 0.0f)
|
|
{
|
|
LOGI("Transferring color from point cloud to mesh...");
|
|
// transfer color from point cloud to mesh
|
|
pcl::search::KdTree<pcl::PointXYZRGBNormal>::Ptr tree (new pcl::search::KdTree<pcl::PointXYZRGBNormal>(true));
|
|
tree->setInputCloud(mergedClouds);
|
|
pcl::PointCloud<pcl::PointXYZRGBNormal>::Ptr coloredCloud(new pcl::PointCloud<pcl::PointXYZRGBNormal>);
|
|
pcl::fromPCLPointCloud2(mesh->cloud, *coloredCloud);
|
|
std::vector<bool> coloredPts(coloredCloud->size());
|
|
for(unsigned int i=0; i<coloredCloud->size(); ++i)
|
|
{
|
|
std::vector<int> kIndices;
|
|
std::vector<float> kDistances;
|
|
pcl::PointXYZRGBNormal pt;
|
|
pt.x = coloredCloud->at(i).x;
|
|
pt.y = coloredCloud->at(i).y;
|
|
pt.z = coloredCloud->at(i).z;
|
|
if(optimizedColorRadius > 0.0f)
|
|
{
|
|
tree->radiusSearch(pt, optimizedColorRadius, kIndices, kDistances);
|
|
}
|
|
else
|
|
{
|
|
tree->nearestKSearch(pt, 1, kIndices, kDistances);
|
|
}
|
|
if(kIndices.size())
|
|
{
|
|
coloredCloud->at(i).r = mergedClouds->at(kIndices[0]).r;
|
|
coloredCloud->at(i).g = mergedClouds->at(kIndices[0]).g;
|
|
coloredCloud->at(i).b = mergedClouds->at(kIndices[0]).b;
|
|
coloredCloud->at(i).a = mergedClouds->at(kIndices[0]).a;
|
|
coloredPts.at(i) = true;
|
|
}
|
|
else
|
|
{
|
|
//white
|
|
coloredCloud->at(i).r = coloredCloud->at(i).g = coloredCloud->at(i).b = 255;
|
|
coloredPts.at(i) = false;
|
|
}
|
|
}
|
|
|
|
// recompute normals and remove polygons with no color
|
|
std::vector<pcl::Vertices> filteredPolygons(optimizedCleanWhitePolygons?mesh->polygons.size():0);
|
|
int oi=0;
|
|
for(unsigned int i=0; i<mesh->polygons.size(); ++i)
|
|
{
|
|
// recompute normals
|
|
pcl::Vertices & v = mesh->polygons[i];
|
|
UASSERT(v.vertices.size()>2);
|
|
Eigen::Vector3f v0(
|
|
coloredCloud->at(v.vertices[1]).x - coloredCloud->at(v.vertices[0]).x,
|
|
coloredCloud->at(v.vertices[1]).y - coloredCloud->at(v.vertices[0]).y,
|
|
coloredCloud->at(v.vertices[1]).z - coloredCloud->at(v.vertices[0]).z);
|
|
int last = v.vertices.size()-1;
|
|
Eigen::Vector3f v1(
|
|
coloredCloud->at(v.vertices[last]).x - coloredCloud->at(v.vertices[0]).x,
|
|
coloredCloud->at(v.vertices[last]).y - coloredCloud->at(v.vertices[0]).y,
|
|
coloredCloud->at(v.vertices[last]).z - coloredCloud->at(v.vertices[0]).z);
|
|
Eigen::Vector3f normal = v0.cross(v1);
|
|
normal.normalize();
|
|
// flat normal (per face)
|
|
for(unsigned int j=0; j<v.vertices.size(); ++j)
|
|
{
|
|
coloredCloud->at(v.vertices[j]).normal_x = normal[0];
|
|
coloredCloud->at(v.vertices[j]).normal_y = normal[1];
|
|
coloredCloud->at(v.vertices[j]).normal_z = normal[2];
|
|
}
|
|
|
|
if(optimizedCleanWhitePolygons)
|
|
{
|
|
bool coloredPolygon = true;
|
|
for(unsigned int j=0; j<mesh->polygons[i].vertices.size(); ++j)
|
|
{
|
|
if(!coloredPts.at(mesh->polygons[i].vertices[j]))
|
|
{
|
|
coloredPolygon = false;
|
|
break;
|
|
}
|
|
}
|
|
if(coloredPolygon)
|
|
{
|
|
filteredPolygons[oi++] = mesh->polygons[i];
|
|
}
|
|
}
|
|
}
|
|
if(optimizedCleanWhitePolygons)
|
|
{
|
|
filteredPolygons.resize(oi);
|
|
mesh->polygons = filteredPolygons;
|
|
}
|
|
|
|
pcl::toPCLPointCloud2(*coloredCloud, mesh->cloud);
|
|
LOGI("Transfering color from point cloud to mesh...done! %fs", timer.ticks());
|
|
}
|
|
else // recompute normals
|
|
{
|
|
pcl::PointCloud<pcl::PointXYZRGBNormal>::Ptr cloud (new pcl::PointCloud<pcl::PointXYZRGBNormal>);
|
|
pcl::fromPCLPointCloud2(mesh->cloud, *cloud);
|
|
|
|
for(unsigned int i=0; i<mesh->polygons.size(); ++i)
|
|
{
|
|
pcl::Vertices & v = mesh->polygons[i];
|
|
UASSERT(v.vertices.size()>2);
|
|
Eigen::Vector3f v0(
|
|
cloud->at(v.vertices[1]).x - cloud->at(v.vertices[0]).x,
|
|
cloud->at(v.vertices[1]).y - cloud->at(v.vertices[0]).y,
|
|
cloud->at(v.vertices[1]).z - cloud->at(v.vertices[0]).z);
|
|
int last = v.vertices.size()-1;
|
|
Eigen::Vector3f v1(
|
|
cloud->at(v.vertices[last]).x - cloud->at(v.vertices[0]).x,
|
|
cloud->at(v.vertices[last]).y - cloud->at(v.vertices[0]).y,
|
|
cloud->at(v.vertices[last]).z - cloud->at(v.vertices[0]).z);
|
|
Eigen::Vector3f normal = v0.cross(v1);
|
|
normal.normalize();
|
|
// flat normal (per face)
|
|
for(unsigned int j=0; j<v.vertices.size(); ++j)
|
|
{
|
|
cloud->at(v.vertices[j]).normal_x = normal[0];
|
|
cloud->at(v.vertices[j]).normal_y = normal[1];
|
|
cloud->at(v.vertices[j]).normal_z = normal[2];
|
|
}
|
|
}
|
|
pcl::toPCLPointCloud2 (*cloud, mesh->cloud);
|
|
}
|
|
polygonMesh = mesh;
|
|
totalPolygons = mesh->polygons.size();
|
|
}
|
|
else
|
|
{
|
|
if(optimizedColorRadius > 0.0f && optimizedCleanWhitePolygons)
|
|
{
|
|
LOGI("Removing polygons too far from the cloud");
|
|
// transfer color from point cloud to mesh
|
|
pcl::search::KdTree<pcl::PointXYZRGBNormal>::Ptr tree (new pcl::search::KdTree<pcl::PointXYZRGBNormal>(true));
|
|
tree->setInputCloud(mergedClouds);
|
|
pcl::PointCloud<pcl::PointXYZ>::Ptr optimizedCloud(new pcl::PointCloud<pcl::PointXYZ>);
|
|
pcl::fromPCLPointCloud2(mesh->cloud, *optimizedCloud);
|
|
std::vector<bool> closePts(optimizedCloud->size());
|
|
for(unsigned int i=0; i<optimizedCloud->size(); ++i)
|
|
{
|
|
std::vector<int> kIndices;
|
|
std::vector<float> kDistances;
|
|
pcl::PointXYZRGBNormal pt;
|
|
pt.x = optimizedCloud->at(i).x;
|
|
pt.y = optimizedCloud->at(i).y;
|
|
pt.z = optimizedCloud->at(i).z;
|
|
tree->radiusSearch(pt, optimizedColorRadius, kIndices, kDistances);
|
|
if(kIndices.size())
|
|
{
|
|
closePts.at(i) = true;
|
|
}
|
|
else
|
|
{
|
|
closePts.at(i) = false;
|
|
}
|
|
}
|
|
|
|
// remove far polygons
|
|
std::vector<pcl::Vertices> filteredPolygons(mesh->polygons.size());
|
|
int oi=0;
|
|
for(unsigned int i=0; i<mesh->polygons.size(); ++i)
|
|
{
|
|
bool keepPolygon = true;
|
|
for(unsigned int j=0; j<mesh->polygons[i].vertices.size(); ++j)
|
|
{
|
|
if(!closePts.at(mesh->polygons[i].vertices[j]))
|
|
{
|
|
keepPolygon = false;
|
|
break;
|
|
}
|
|
}
|
|
if(keepPolygon)
|
|
{
|
|
filteredPolygons[oi++] = mesh->polygons[i];
|
|
}
|
|
}
|
|
filteredPolygons.resize(oi);
|
|
mesh->polygons = filteredPolygons;
|
|
|
|
LOGI("Removing polygons too far from the cloud...done! %fs", timer.ticks());
|
|
}
|
|
|
|
LOGI("Texturing...");
|
|
textureMesh = rtabmap::util3d::createTextureMesh(
|
|
mesh,
|
|
cameraPoses,
|
|
cameraModels,
|
|
maxTextureDistance);
|
|
LOGI("Texturing... done! %fs", timer.ticks());
|
|
|
|
// Remove occluded polygons (polygons with no texture)
|
|
if(textureMesh->tex_coordinates.size() && optimizedCleanWhitePolygons)
|
|
{
|
|
LOGI("Cleanup mesh...");
|
|
|
|
// assume last texture is the occluded texture
|
|
textureMesh->tex_coordinates.pop_back();
|
|
textureMesh->tex_polygons.pop_back();
|
|
textureMesh->tex_materials.pop_back();
|
|
|
|
if(minClusterSize_>0)
|
|
{
|
|
LOGI("Filter small polygon clusters...");
|
|
|
|
// concatenate all polygons
|
|
int totalSize = 0;
|
|
for(unsigned int t=0; t<textureMesh->tex_polygons.size(); ++t)
|
|
{
|
|
totalSize+=textureMesh->tex_polygons[t].size();
|
|
}
|
|
std::vector<pcl::Vertices> allPolygons(totalSize);
|
|
int oi=0;
|
|
for(unsigned int t=0; t<textureMesh->tex_polygons.size(); ++t)
|
|
{
|
|
for(unsigned int i=0; i<textureMesh->tex_polygons[t].size(); ++i)
|
|
{
|
|
allPolygons[oi++] = textureMesh->tex_polygons[t][i];
|
|
}
|
|
}
|
|
|
|
// filter polygons
|
|
std::vector<std::set<int> > neighbors;
|
|
std::vector<std::set<int> > vertexToPolygons;
|
|
rtabmap::util3d::createPolygonIndexes(allPolygons,
|
|
textureMesh->cloud.data.size()/textureMesh->cloud.point_step,
|
|
neighbors,
|
|
vertexToPolygons);
|
|
std::list<std::list<int> > clusters = rtabmap::util3d::clusterPolygons(
|
|
neighbors,
|
|
minClusterSize_);
|
|
|
|
std::set<int> validPolygons;
|
|
for(std::list<std::list<int> >::iterator kter=clusters.begin(); kter!=clusters.end(); ++kter)
|
|
{
|
|
for(std::list<int>::iterator jter=kter->begin(); jter!=kter->end(); ++jter)
|
|
{
|
|
validPolygons.insert(*jter);
|
|
}
|
|
}
|
|
|
|
// for each texture
|
|
unsigned int allPolygonsIndex = 0;
|
|
for(unsigned int t=0; t<textureMesh->tex_polygons.size(); ++t)
|
|
{
|
|
std::vector<pcl::Vertices> filteredPolygons(textureMesh->tex_polygons[t].size());
|
|
#if PCL_VERSION_COMPARE(>=, 1, 8, 0)
|
|
std::vector<Eigen::Vector2f, Eigen::aligned_allocator<Eigen::Vector2f> > filteredCoordinates(textureMesh->tex_coordinates[t].size());
|
|
#else
|
|
std::vector<Eigen::Vector2f> filteredCoordinates(textureMesh->tex_coordinates[t].size());
|
|
#endif
|
|
int oi=0;
|
|
unsigned int polygonSize = 0;
|
|
if(textureMesh->tex_polygons[t].size())
|
|
{
|
|
UASSERT(allPolygonsIndex < allPolygons.size());
|
|
|
|
polygonSize = textureMesh->tex_polygons[t][0].vertices.size();
|
|
|
|
UASSERT(filteredCoordinates.size() == textureMesh->tex_polygons[t].size()*polygonSize);
|
|
for(unsigned int i=0; i<textureMesh->tex_polygons[t].size(); ++i)
|
|
{
|
|
if(validPolygons.find(allPolygonsIndex) != validPolygons.end())
|
|
{
|
|
filteredPolygons[oi] = textureMesh->tex_polygons[t].at(i);
|
|
for(unsigned int j=0; j<polygonSize; ++j)
|
|
{
|
|
filteredCoordinates[oi*polygonSize + j] = textureMesh->tex_coordinates[t][i*polygonSize + j];
|
|
}
|
|
++oi;
|
|
}
|
|
++allPolygonsIndex;
|
|
}
|
|
filteredPolygons.resize(oi);
|
|
filteredCoordinates.resize(oi*polygonSize);
|
|
textureMesh->tex_polygons[t] = filteredPolygons;
|
|
textureMesh->tex_coordinates[t] = filteredCoordinates;
|
|
}
|
|
}
|
|
|
|
LOGI("Filtered %d polygons.", (int)(allPolygons.size()-validPolygons.size()));
|
|
}
|
|
|
|
for(unsigned int t=0; t<textureMesh->tex_polygons.size(); ++t)
|
|
{
|
|
totalPolygons+=textureMesh->tex_polygons[t].size();
|
|
}
|
|
|
|
LOGI("Cleanup mesh... done! %fs (total polygons=%d)", timer.ticks(), totalPolygons);
|
|
}
|
|
else
|
|
{
|
|
for(unsigned int t=0; t<textureMesh->tex_polygons.size(); ++t)
|
|
{
|
|
totalPolygons+=textureMesh->tex_polygons[t].size();
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
else // organized meshes
|
|
{
|
|
pcl::PointCloud<pcl::PointXYZRGBNormal>::Ptr mergedClouds(new pcl::PointCloud<pcl::PointXYZRGBNormal>);
|
|
|
|
if(textureSize > 0)
|
|
{
|
|
textureMesh->tex_materials.resize(poses.size());
|
|
textureMesh->tex_polygons.resize(poses.size());
|
|
textureMesh->tex_coordinates.resize(poses.size());
|
|
}
|
|
|
|
int polygonsStep = 0;
|
|
int oi = 0;
|
|
for(std::map<int, rtabmap::Transform>::iterator iter=poses.begin();
|
|
iter!= poses.end();
|
|
++iter)
|
|
{
|
|
LOGI("Assembling cloud %d (total=%d)...", iter->first, (int)poses.size());
|
|
|
|
std::map<int, Mesh>::iterator jter = createdMeshes_.find(iter->first);
|
|
pcl::PointCloud<pcl::PointXYZRGB>::Ptr cloud(new pcl::PointCloud<pcl::PointXYZRGB>);
|
|
std::vector<pcl::Vertices> polygons;
|
|
float gain = 1.0f;
|
|
if(jter != createdMeshes_.end())
|
|
{
|
|
cloud = jter->second.cloud;
|
|
polygons= jter->second.polygons;
|
|
if(cloud->size() && polygons.size() == 0)
|
|
{
|
|
polygons = rtabmap::util3d::organizedFastMesh(cloud, meshAngleToleranceDeg_*M_PI/180.0, false, meshTrianglePix_);
|
|
}
|
|
gain = jter->second.gain;
|
|
}
|
|
else
|
|
{
|
|
rtabmap::SensorData data = rtabmap_->getMemory()->getNodeData(iter->first, true);
|
|
if(!data.imageRaw().empty() && !data.depthRaw().empty() && data.cameraModels().size() == 1)
|
|
{
|
|
cloud = rtabmap::util3d::cloudRGBFromSensorData(data, meshDecimation_, maxCloudDepth_, 0);
|
|
polygons = rtabmap::util3d::organizedFastMesh(cloud, meshAngleToleranceDeg_*M_PI/180.0, false, meshTrianglePix_);
|
|
}
|
|
}
|
|
|
|
if(cloud->size() && polygons.size())
|
|
{
|
|
// Convert organized to dense cloud
|
|
pcl::PointCloud<pcl::PointXYZRGB>::Ptr outputCloud(new pcl::PointCloud<pcl::PointXYZRGB>);
|
|
std::vector<pcl::Vertices> outputPolygons;
|
|
std::vector<int> denseToOrganizedIndices = rtabmap::util3d::filterNaNPointsFromMesh(*cloud, polygons, *outputCloud, outputPolygons);
|
|
|
|
pcl::PointCloud<pcl::Normal>::Ptr normals = rtabmap::util3d::computeNormals(outputCloud, normalK);
|
|
|
|
pcl::PointCloud<pcl::PointXYZRGBNormal>::Ptr cloudWithNormals(new pcl::PointCloud<pcl::PointXYZRGBNormal>);
|
|
pcl::concatenateFields(*outputCloud, *normals, *cloudWithNormals);
|
|
|
|
UASSERT(outputPolygons.size());
|
|
|
|
totalPolygons+=outputPolygons.size();
|
|
|
|
if(textureSize == 0)
|
|
{
|
|
// colored mesh
|
|
cloudWithNormals = rtabmap::util3d::transformPointCloud(cloudWithNormals, iter->second);
|
|
|
|
if(gain != 1.0f)
|
|
{
|
|
for(unsigned int i=0; i<cloudWithNormals->size(); ++i)
|
|
{
|
|
pcl::PointXYZRGBNormal & pt = cloudWithNormals->at(i);
|
|
pt.r = uchar(std::max(0.0, std::min(255.0, double(pt.r) * gain)));
|
|
pt.g = uchar(std::max(0.0, std::min(255.0, double(pt.g) * gain)));
|
|
pt.b = uchar(std::max(0.0, std::min(255.0, double(pt.b) * gain)));
|
|
}
|
|
}
|
|
|
|
if(mergedClouds->size() == 0)
|
|
{
|
|
*mergedClouds = *cloudWithNormals;
|
|
polygonMesh->polygons = outputPolygons;
|
|
}
|
|
else
|
|
{
|
|
rtabmap::util3d::appendMesh(*mergedClouds, polygonMesh->polygons, *cloudWithNormals, outputPolygons);
|
|
}
|
|
}
|
|
else
|
|
{
|
|
// texture mesh
|
|
unsigned int polygonSize = outputPolygons.front().vertices.size();
|
|
textureMesh->tex_polygons[oi].resize(outputPolygons.size());
|
|
textureMesh->tex_coordinates[oi].resize(outputPolygons.size() * polygonSize);
|
|
for(unsigned int j=0; j<outputPolygons.size(); ++j)
|
|
{
|
|
pcl::Vertices vertices = outputPolygons[j];
|
|
UASSERT(polygonSize == vertices.vertices.size());
|
|
for(unsigned int k=0; k<vertices.vertices.size(); ++k)
|
|
{
|
|
//uv
|
|
UASSERT(vertices.vertices[k] < denseToOrganizedIndices.size());
|
|
int originalVertex = denseToOrganizedIndices[vertices.vertices[k]];
|
|
textureMesh->tex_coordinates[oi][j*vertices.vertices.size()+k] = Eigen::Vector2f(
|
|
float(originalVertex % cloud->width) / float(cloud->width), // u
|
|
float(cloud->height - originalVertex / cloud->width) / float(cloud->height)); // v
|
|
|
|
vertices.vertices[k] += polygonsStep;
|
|
}
|
|
textureMesh->tex_polygons[oi][j] = vertices;
|
|
|
|
}
|
|
polygonsStep += outputCloud->size();
|
|
|
|
pcl::PointCloud<pcl::PointXYZRGBNormal>::Ptr transformedCloud = rtabmap::util3d::transformPointCloud(cloudWithNormals, iter->second);
|
|
if(mergedClouds->size() == 0)
|
|
{
|
|
*mergedClouds = *transformedCloud;
|
|
}
|
|
else
|
|
{
|
|
*mergedClouds += *transformedCloud;
|
|
}
|
|
|
|
textureMesh->tex_materials[oi].tex_illum = 1;
|
|
textureMesh->tex_materials[oi].tex_name = uFormat("material_%d", iter->first);
|
|
textureMesh->tex_materials[oi].tex_file = uNumber2Str(iter->first);
|
|
++oi;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
UERROR("Mesh not found for mesh %d", iter->first);
|
|
}
|
|
}
|
|
if(textureSize == 0)
|
|
{
|
|
if(mergedClouds->size())
|
|
{
|
|
pcl::toPCLPointCloud2(*mergedClouds, polygonMesh->cloud);
|
|
}
|
|
else
|
|
{
|
|
polygonMesh->polygons.clear();
|
|
}
|
|
}
|
|
else
|
|
{
|
|
textureMesh->tex_materials.resize(oi);
|
|
textureMesh->tex_polygons.resize(oi);
|
|
|
|
if(mergedClouds->size())
|
|
{
|
|
pcl::toPCLPointCloud2(*mergedClouds, textureMesh->cloud);
|
|
}
|
|
}
|
|
}
|
|
|
|
// end optimized or organized
|
|
|
|
if(textureSize>0 && totalPolygons && textureMesh->tex_materials.size())
|
|
{
|
|
LOGI("Merging %d textures...", (int)textureMesh->tex_materials.size());
|
|
globalTexture = mergeTextures(*textureMesh, textureSize);
|
|
|
|
std::string baseName = uSplit(UFile::getName(filePath), '.').front();
|
|
std::string textureDirectory = UDirectory::getDir(filePath);
|
|
std::string fullPath = textureDirectory+UDirectory::separator()+baseName+".jpg";
|
|
textureMesh->tex_materials[0].tex_file = baseName+".jpg";
|
|
LOGI("Saving texture to %s.", fullPath.c_str());
|
|
if(!cv::imwrite(fullPath, globalTexture))
|
|
{
|
|
LOGI("Failed saving %s!", fullPath.c_str());
|
|
}
|
|
else
|
|
{
|
|
LOGI("Saved %s (%d bytes).", fullPath.c_str(), globalTexture.total()*globalTexture.channels());
|
|
}
|
|
}
|
|
}
|
|
if(totalPolygons)
|
|
{
|
|
if(textureSize == 0)
|
|
{
|
|
UASSERT((int)polygonMesh->polygons.size() == totalPolygons);
|
|
if(polygonMesh->polygons.size())
|
|
{
|
|
LOGI("Saving ply (%d vertices, %d polygons) to %s.", (int)polygonMesh->cloud.data.size()/polygonMesh->cloud.point_step, totalPolygons, filePath.c_str());
|
|
success = pcl::io::savePLYFile(filePath, *polygonMesh) == 0;
|
|
if(success)
|
|
{
|
|
UINFO("Saved ply to %s!", filePath.c_str());
|
|
exportedMesh_.reset(new pcl::TextureMesh);
|
|
exportedMesh_->cloud = polygonMesh->cloud;
|
|
exportedMesh_->tex_polygons.push_back(polygonMesh->polygons);
|
|
}
|
|
else
|
|
{
|
|
UERROR("Failed saving ply to %s!", filePath.c_str());
|
|
}
|
|
}
|
|
}
|
|
else if(textureMesh->tex_materials.size())
|
|
{
|
|
UASSERT(textureMesh->tex_polygons.size() && (int)textureMesh->tex_polygons[0].size() == totalPolygons);
|
|
LOGI("Saving obj (%d vertices, %d polygons) to %s.", (int)textureMesh->cloud.data.size()/textureMesh->cloud.point_step, totalPolygons, filePath.c_str());
|
|
success = pcl::io::saveOBJFile(filePath, *textureMesh) == 0;
|
|
if(success)
|
|
{
|
|
LOGI("Saved obj to %s!", filePath.c_str());
|
|
exportedMesh_ = textureMesh;
|
|
exportedTexture_ = globalTexture;
|
|
}
|
|
else
|
|
{
|
|
UERROR("Failed saving obj to %s!", filePath.c_str());
|
|
}
|
|
}
|
|
else
|
|
{
|
|
UERROR("Failed exporting obj to %s! There are no textures!", filePath.c_str());
|
|
}
|
|
}
|
|
else
|
|
{
|
|
UERROR("Failed exporting to %s! There are no polygons!", filePath.c_str());
|
|
}
|
|
}
|
|
else // Point cloud
|
|
{
|
|
pcl::PointCloud<pcl::PointXYZRGB>::Ptr mergedClouds(new pcl::PointCloud<pcl::PointXYZRGB>);
|
|
for(std::map<int, rtabmap::Transform>::iterator iter=poses.begin();
|
|
iter!= poses.end();
|
|
++iter)
|
|
{
|
|
std::map<int, Mesh>::iterator jter=createdMeshes_.find(iter->first);
|
|
pcl::PointCloud<pcl::PointXYZRGB>::Ptr cloud(new pcl::PointCloud<pcl::PointXYZRGB>);
|
|
pcl::IndicesPtr indices(new std::vector<int>);
|
|
float gain = 1.0f;
|
|
if(regenerateCloud)
|
|
{
|
|
if(jter != createdMeshes_.end())
|
|
{
|
|
gain = jter->second.gain;
|
|
}
|
|
rtabmap::SensorData data = rtabmap_->getMemory()->getNodeData(iter->first, true);
|
|
if(!data.imageRaw().empty() && !data.depthRaw().empty())
|
|
{
|
|
// full resolution
|
|
cloud = rtabmap::util3d::cloudRGBFromSensorData(data, 1, maxCloudDepth_, 0, indices.get());
|
|
}
|
|
}
|
|
else
|
|
{
|
|
if(jter != createdMeshes_.end())
|
|
{
|
|
cloud = jter->second.cloud;
|
|
indices = jter->second.indices;
|
|
gain = jter->second.gain;
|
|
}
|
|
else
|
|
{
|
|
rtabmap::SensorData data = rtabmap_->getMemory()->getNodeData(iter->first, true);
|
|
if(!data.imageRaw().empty() && !data.depthRaw().empty())
|
|
{
|
|
cloud = rtabmap::util3d::cloudRGBFromSensorData(data, meshDecimation_, maxCloudDepth_, 0, indices.get());
|
|
}
|
|
}
|
|
}
|
|
if(cloud->size() && indices->size())
|
|
{
|
|
// Convert organized to dense cloud
|
|
pcl::PointCloud<pcl::PointXYZRGB>::Ptr transformedCloud(new pcl::PointCloud<pcl::PointXYZRGB>);
|
|
if(cloudVoxelSize > 0.0f)
|
|
{
|
|
transformedCloud = rtabmap::util3d::voxelize(cloud, indices, cloudVoxelSize);
|
|
transformedCloud = rtabmap::util3d::transformPointCloud(transformedCloud, iter->second);
|
|
}
|
|
else
|
|
{
|
|
// it looks like that using only transformPointCloud with indices
|
|
// flushes the colors, so we should extract points before... maybe a too old PCL version
|
|
pcl::copyPointCloud(*cloud, *indices, *transformedCloud);
|
|
transformedCloud = rtabmap::util3d::transformPointCloud(transformedCloud, iter->second);
|
|
}
|
|
|
|
if(gain != 1.0f)
|
|
{
|
|
//LOGD("cloud %d, gain=%f", iter->first, gain);
|
|
for(unsigned int i=0; i<transformedCloud->size(); ++i)
|
|
{
|
|
pcl::PointXYZRGB & pt = transformedCloud->at(i);
|
|
//LOGI("color %d = %d %d %d", i, (int)pt.r, (int)pt.g, (int)pt.b);
|
|
pt.r = uchar(std::max(0.0, std::min(255.0, double(pt.r) * gain)));
|
|
pt.g = uchar(std::max(0.0, std::min(255.0, double(pt.g) * gain)));
|
|
pt.b = uchar(std::max(0.0, std::min(255.0, double(pt.b) * gain)));
|
|
|
|
}
|
|
}
|
|
|
|
if(mergedClouds->size() == 0)
|
|
{
|
|
*mergedClouds = *transformedCloud;
|
|
}
|
|
else
|
|
{
|
|
*mergedClouds += *transformedCloud;
|
|
}
|
|
}
|
|
}
|
|
|
|
if(mergedClouds->size())
|
|
{
|
|
if(cloudVoxelSize > 0.0f)
|
|
{
|
|
mergedClouds = rtabmap::util3d::voxelize(mergedClouds, cloudVoxelSize);
|
|
}
|
|
|
|
pcl::PolygonMesh mesh;
|
|
pcl::toPCLPointCloud2(*mergedClouds, mesh.cloud);
|
|
|
|
LOGI("Saving ply (%d points) to %s.", (int)mergedClouds->size(), filePath.c_str());
|
|
success = pcl::io::savePLYFileBinary(filePath, mesh) == 0;
|
|
if(success)
|
|
{
|
|
LOGI("Saved ply to %s!", filePath.c_str());
|
|
mergedClouds->clear();
|
|
exportedMesh_.reset(new pcl::TextureMesh);
|
|
exportedMesh_->cloud = mesh.cloud;
|
|
}
|
|
else
|
|
{
|
|
UERROR("Failed saving ply to %s!", filePath.c_str());
|
|
}
|
|
}
|
|
}
|
|
|
|
if(blockRendering)
|
|
{
|
|
renderingMutex_.unlock();
|
|
}
|
|
}
|
|
catch (std::exception & e)
|
|
{
|
|
UERROR("Out of memory! %s", e.what());
|
|
|
|
if(blockRendering)
|
|
{
|
|
renderingMutex_.unlock();
|
|
}
|
|
|
|
success = false;
|
|
}
|
|
|
|
return success;
|
|
}
|
|
|
|
bool RTABMapApp::postExportation(bool visualize)
|
|
{
|
|
LOGI("postExportation(visualize=%d)", visualize?1:0);
|
|
if(visualize && exportedMesh_->cloud.data.size())
|
|
{
|
|
boost::mutex::scoped_lock lock(renderingMutex_);
|
|
visualizingMesh_ = true;
|
|
exportedMeshUpdated_ = true;
|
|
}
|
|
else
|
|
{
|
|
exportedMesh_.reset(new pcl::TextureMesh);
|
|
exportedTexture_ = cv::Mat();
|
|
exportedMeshUpdated_ = false;
|
|
visualizingMesh_ = false;
|
|
}
|
|
|
|
return visualizingMesh_;
|
|
}
|
|
|
|
int RTABMapApp::postProcessing(int approach)
|
|
{
|
|
LOGI("postProcessing(%d)", approach);
|
|
int returnedValue = 0;
|
|
if(rtabmap_)
|
|
{
|
|
std::map<int, rtabmap::Transform> poses;
|
|
std::multimap<int, rtabmap::Link> links;
|
|
|
|
// detect more loop closures
|
|
if(approach == -1 || approach == 2)
|
|
{
|
|
returnedValue = rtabmap_->detectMoreLoopClosures(1.0f, M_PI/6.0f, approach == -1?5:1);
|
|
}
|
|
|
|
// graph optimization
|
|
if(returnedValue >=0)
|
|
{
|
|
if (approach == 1)
|
|
{
|
|
if(rtabmap::Optimizer::isAvailable(rtabmap::Optimizer::kTypeG2O))
|
|
{
|
|
std::map<int, rtabmap::Signature> signatures;
|
|
rtabmap_->getGraph(poses, links, true, true, &signatures);
|
|
|
|
rtabmap::ParametersMap param;
|
|
param.insert(rtabmap::ParametersPair(rtabmap::Parameters::kOptimizerIterations(), "30"));
|
|
param.insert(rtabmap::ParametersPair(rtabmap::Parameters::kOptimizerEpsilon(), "0"));
|
|
rtabmap::Optimizer * sba = rtabmap::Optimizer::create(rtabmap::Optimizer::kTypeG2O, param);
|
|
poses = sba->optimizeBA(poses.rbegin()->first, poses, links, signatures);
|
|
delete sba;
|
|
}
|
|
else
|
|
{
|
|
UERROR("g2o not available!");
|
|
}
|
|
}
|
|
else if(approach!=4 && approach!=5 && approach != 7)
|
|
{
|
|
// simple graph optmimization
|
|
rtabmap_->getGraph(poses, links, true, true);
|
|
}
|
|
}
|
|
|
|
if(poses.size())
|
|
{
|
|
boost::mutex::scoped_lock lock(rtabmapMutex_);
|
|
rtabmap::Statistics stats = rtabmap_->getStatistics();
|
|
stats.setPoses(poses);
|
|
stats.setConstraints(links);
|
|
|
|
LOGI("PostProcessing, sending rtabmap event to update graph...");
|
|
rtabmapEvents_.push_back(stats);
|
|
|
|
rtabmap_->setOptimizedPoses(poses);
|
|
}
|
|
else if(approach!=4 && approach!=5 && approach != 7)
|
|
{
|
|
returnedValue = -1;
|
|
}
|
|
|
|
if(returnedValue >=0)
|
|
{
|
|
boost::mutex::scoped_lock lock(renderingMutex_);
|
|
// filter polygons
|
|
if(approach == 4)
|
|
{
|
|
filterPolygonsOnNextRender_ = true;
|
|
}
|
|
|
|
// gain compensation
|
|
if(approach == -1 || approach == 5 || approach == 6)
|
|
{
|
|
gainCompensationOnNextRender_ = approach == 6 ? 2 : 1; // 2 = full, 1 = fast
|
|
}
|
|
|
|
// bilateral filtering
|
|
if(approach == 7)
|
|
{
|
|
bilateralFilteringOnNextRender_ = true;
|
|
}
|
|
}
|
|
}
|
|
|
|
return returnedValue;
|
|
}
|
|
|
|
void RTABMapApp::handleEvent(UEvent * event)
|
|
{
|
|
if(camera_ && camera_->isRunning())
|
|
{
|
|
// called from events manager thread, so protect the data
|
|
if(event->getClassName().compare("OdometryEvent") == 0)
|
|
{
|
|
LOGI("Received OdometryEvent!");
|
|
if(odomMutex_.try_lock())
|
|
{
|
|
odomEvents_.clear();
|
|
if(camera_->isRunning())
|
|
{
|
|
odomEvents_.push_back(*((rtabmap::OdometryEvent*)(event)));
|
|
}
|
|
odomMutex_.unlock();
|
|
}
|
|
}
|
|
if(status_.first == rtabmap::RtabmapEventInit::kInitialized &&
|
|
event->getClassName().compare("RtabmapEvent") == 0)
|
|
{
|
|
LOGI("Received RtabmapEvent initialized event!");
|
|
if(camera_->isRunning())
|
|
{
|
|
boost::mutex::scoped_lock lock(rtabmapMutex_);
|
|
rtabmapEvents_.push_back(((rtabmap::RtabmapEvent*)event)->getStats());
|
|
}
|
|
}
|
|
}
|
|
|
|
if(event->getClassName().compare("PoseEvent") == 0)
|
|
{
|
|
if(poseMutex_.try_lock())
|
|
{
|
|
poseEvents_.clear();
|
|
poseEvents_.push_back(((rtabmap::PoseEvent*)event)->pose());
|
|
poseMutex_.unlock();
|
|
}
|
|
}
|
|
|
|
if(event->getClassName().compare("CameraTangoEvent") == 0)
|
|
{
|
|
rtabmap::CameraTangoEvent * tangoEvent = (rtabmap::CameraTangoEvent*)event;
|
|
|
|
// Call JAVA callback with tango event msg
|
|
bool success = false;
|
|
if(jvm && RTABMapActivity)
|
|
{
|
|
JNIEnv *env = 0;
|
|
jint rs = jvm->AttachCurrentThread(&env, NULL);
|
|
if(rs == JNI_OK && env)
|
|
{
|
|
jclass clazz = env->GetObjectClass(RTABMapActivity);
|
|
if(clazz)
|
|
{
|
|
jmethodID methodID = env->GetMethodID(clazz, "tangoEventCallback", "(ILjava/lang/String;Ljava/lang/String;)V" );
|
|
if(methodID)
|
|
{
|
|
env->CallVoidMethod(RTABMapActivity, methodID,
|
|
tangoEvent->type(),
|
|
env->NewStringUTF(tangoEvent->key().c_str()),
|
|
env->NewStringUTF(tangoEvent->value().c_str()));
|
|
success = true;
|
|
}
|
|
}
|
|
}
|
|
jvm->DetachCurrentThread();
|
|
}
|
|
if(!success)
|
|
{
|
|
UERROR("Failed to call RTABMapActivity::tangoEventCallback");
|
|
}
|
|
}
|
|
|
|
if(event->getClassName().compare("RtabmapEventInit") == 0)
|
|
{
|
|
LOGI("Received RtabmapEventInit!");
|
|
status_.first = ((rtabmap::RtabmapEventInit*)event)->getStatus();
|
|
status_.second = ((rtabmap::RtabmapEventInit*)event)->getInfo();
|
|
|
|
// Call JAVA callback with init msg
|
|
bool success = false;
|
|
if(jvm && RTABMapActivity)
|
|
{
|
|
JNIEnv *env = 0;
|
|
jint rs = jvm->AttachCurrentThread(&env, NULL);
|
|
if(rs == JNI_OK && env)
|
|
{
|
|
jclass clazz = env->GetObjectClass(RTABMapActivity);
|
|
if(clazz)
|
|
{
|
|
jmethodID methodID = env->GetMethodID(clazz, "rtabmapInitEventCallback", "(ILjava/lang/String;)V" );
|
|
if(methodID)
|
|
{
|
|
env->CallVoidMethod(RTABMapActivity, methodID,
|
|
status_.first,
|
|
env->NewStringUTF(status_.second.c_str()));
|
|
success = true;
|
|
}
|
|
}
|
|
}
|
|
jvm->DetachCurrentThread();
|
|
}
|
|
if(!success)
|
|
{
|
|
UERROR("Failed to call RTABMapActivity::rtabmapInitEventsCallback");
|
|
}
|
|
}
|
|
|
|
if(event->getClassName().compare("PostRenderEvent") == 0)
|
|
{
|
|
LOGI("Received PostRenderEvent!");
|
|
const rtabmap::Statistics & stats = ((PostRenderEvent*)event)->getStats();
|
|
int nodes = (int)uValue(stats.data(), rtabmap::Statistics::kMemoryWorking_memory_size(), 0.0f) +
|
|
uValue(stats.data(), rtabmap::Statistics::kMemoryShort_time_memory_size(), 0.0f);
|
|
int words = (int)uValue(stats.data(), rtabmap::Statistics::kKeypointDictionary_size(), 0.0f);
|
|
float updateTime = uValue(stats.data(), rtabmap::Statistics::kTimingTotal(), 0.0f);
|
|
int loopClosureId = stats.loopClosureId()>0?stats.loopClosureId():stats.proximityDetectionId()>0?stats.proximityDetectionId():0;
|
|
int highestHypId = (int)uValue(stats.data(), rtabmap::Statistics::kLoopHighest_hypothesis_id(), 0.0f);
|
|
int databaseMemoryUsed = (int)uValue(stats.data(), rtabmap::Statistics::kMemoryDatabase_memory_used(), 0.0f);
|
|
int inliers = (int)uValue(stats.data(), rtabmap::Statistics::kLoopVisual_inliers(), 0.0f);
|
|
int matches = (int)uValue(stats.data(), rtabmap::Statistics::kLoopVisual_matches(), 0.0f);
|
|
int rejected = (int)uValue(stats.data(), rtabmap::Statistics::kLoopRejectedHypothesis(), 0.0f);
|
|
float optimizationMaxError = uValue(stats.data(), rtabmap::Statistics::kLoopOptimization_max_error(), 0.0f);
|
|
float rehearsalValue = uValue(stats.data(), rtabmap::Statistics::kMemoryRehearsal_sim(), 0.0f);
|
|
int featuresExtracted = stats.getSignatures().size()?stats.getSignatures().rbegin()->second.getWords().size():0;
|
|
float hypothesis = uValue(stats.data(), rtabmap::Statistics::kLoopHighest_hypothesis_value(), 0.0f);
|
|
|
|
// Call JAVA callback with some stats
|
|
UINFO("Send statistics to GUI");
|
|
bool success = false;
|
|
if(jvm && RTABMapActivity)
|
|
{
|
|
JNIEnv *env = 0;
|
|
jint rs = jvm->AttachCurrentThread(&env, NULL);
|
|
if(rs == JNI_OK && env)
|
|
{
|
|
jclass clazz = env->GetObjectClass(RTABMapActivity);
|
|
if(clazz)
|
|
{
|
|
jmethodID methodID = env->GetMethodID(clazz, "updateStatsCallback", "(IIIIFIIIIIIIFIFIFF)V" );
|
|
if(methodID)
|
|
{
|
|
env->CallVoidMethod(RTABMapActivity, methodID,
|
|
nodes,
|
|
words,
|
|
totalPoints_,
|
|
totalPolygons_,
|
|
updateTime,
|
|
loopClosureId,
|
|
highestHypId,
|
|
(int)((processMemoryUsedBytes+processGPUMemoryUsedBytes)/(1024*1024)),
|
|
databaseMemoryUsed,
|
|
inliers,
|
|
matches,
|
|
featuresExtracted,
|
|
hypothesis,
|
|
lastDrawnCloudsCount_,
|
|
renderingTime_>0.0f?1.0f/renderingTime_:0.0f,
|
|
rejected,
|
|
rehearsalValue,
|
|
optimizationMaxError);
|
|
success = true;
|
|
}
|
|
}
|
|
}
|
|
jvm->DetachCurrentThread();
|
|
}
|
|
if(!success)
|
|
{
|
|
UERROR("Failed to call RTABMapActivity::updateStatsCallback");
|
|
}
|
|
renderingTime_ = 0.0f;
|
|
}
|
|
}
|
|
|