i/rtabmap
1
0
Fork 0
mirror of https://github.com/introlab/rtabmap.git synced 2026-09-02 09:30:25 +08:00
Code Issues Packages Projects Releases Wiki Activity

0.19.2: Refactored SensorData interface. DBReader: Fixed GPS not published. #345: both g2o and gtsam working with GPS. g2o: added gravity edges.

Browse Source
This commit is contained in:
matlabbe
2019-04-09 20:05:24 -04:00
parent e7b3a7735d
commit 77ae8e108a
24 changed files with 878 additions and 480 deletions
Download Patch File Download Diff File Expand all files Collapse all files

57
corelib/src/CameraThread.cpp
View File

@@ -176,7 +176,7 @@ void CameraThread::postUpdate(SensorData * dataPtr, CameraInfo * info) const
SensorData & data = *dataPtr;
if(_colorOnly && !data.depthRaw().empty())
{
data.setDepthOrRightRaw(cv::Mat());
data.setRGBDImage(data.imageRaw(), cv::Mat(), data.cameraModels());
}
if(_distortionModel && !data.depthRaw().empty())
@@ -187,7 +187,7 @@ void CameraThread::postUpdate(SensorData * dataPtr, CameraInfo * info) const
{
cv::Mat depth = data.depthRaw().clone();// make sure we are not modifying data in cached signatures.
_distortionModel->undistort(depth);
data.setDepthOrRightRaw(depth);
data.setRGBDImage(data.imageRaw(), depth, data.cameraModels());
}
else
{
@@ -201,7 +201,7 @@ void CameraThread::postUpdate(SensorData * dataPtr, CameraInfo * info) const
if(_bilateralFiltering && !data.depthRaw().empty())
{
UTimer timer;
data.setDepthOrRightRaw(util2d::fastBilateralFiltering(data.depthRaw(), _bilateralSigmaS, _bilateralSigmaR));
data.setRGBDImage(data.imageRaw(), util2d::fastBilateralFiltering(data.depthRaw(), _bilateralSigmaS, _bilateralSigmaR), data.cameraModels());
if(info) info->timeBilateralFiltering = timer.ticks();
}
@@ -217,8 +217,8 @@ void CameraThread::postUpdate(SensorData * dataPtr, CameraInfo * info) const
}
else
{
data.setImageRaw(util2d::decimate(data.imageRaw(), _imageDecimation));
data.setDepthOrRightRaw(util2d::decimate(data.depthOrRightRaw(), _imageDecimation));
cv::Mat image = util2d::decimate(data.imageRaw(), _imageDecimation);
cv::Mat depthOrRight = util2d::decimate(data.depthOrRightRaw(), _imageDecimation);
std::vector<CameraModel> models = data.cameraModels();
for(unsigned int i=0; i<models.size(); ++i)
{
@@ -227,12 +227,18 @@ void CameraThread::postUpdate(SensorData * dataPtr, CameraInfo * info) const
models[i] = models[i].scaled(1.0/double(_imageDecimation));
}
}
data.setCameraModels(models);
StereoCameraModel stereoModel = data.stereoCameraModel();
if(stereoModel.isValidForProjection())
if(!models.empty())
{
stereoModel.scale(1.0/double(_imageDecimation));
data.setStereoCameraModel(stereoModel);
data.setRGBDImage(image, depthOrRight, models);
}
else
{
StereoCameraModel stereoModel = data.stereoCameraModel();
if(stereoModel.isValidForProjection())
{
stereoModel.scale(1.0/double(_imageDecimation));
}
data.setStereoImage(image, depthOrRight, stereoModel);
}
}
if(info) info->timeImageDecimation = timer.ticks();
@@ -243,11 +249,12 @@ void CameraThread::postUpdate(SensorData * dataPtr, CameraInfo * info) const
UTimer timer;
cv::Mat tmpRgb;
cv::flip(data.imageRaw(), tmpRgb, 1);
data.setImageRaw(tmpRgb);
UASSERT_MSG(data.cameraModels().size() <= 1 && !data.stereoCameraModel().isValidForProjection(), "Only single RGBD cameras are supported for mirroring.");
if(data.cameraModels().size() && data.cameraModels()[0].cx())
CameraModel tmpModel = data.cameraModels()[0];
if(data.cameraModels()[0].cx())
{
CameraModel tmpModel(
tmpModel = CameraModel(
data.cameraModels()[0].fx(),
data.cameraModels()[0].fy(),
float(data.imageRaw().cols) - data.cameraModels()[0].cx(),
@@ -255,14 +262,13 @@ void CameraThread::postUpdate(SensorData * dataPtr, CameraInfo * info) const
data.cameraModels()[0].localTransform(),
data.cameraModels()[0].Tx(),
data.cameraModels()[0].imageSize());
data.setCameraModel(tmpModel);
}
cv::Mat tmpDepth = data.depthOrRightRaw();
if(!data.depthRaw().empty())
{
cv::Mat tmpDepth;
cv::flip(data.depthRaw(), tmpDepth, 1);
data.setDepthOrRightRaw(tmpDepth);
}
data.setRGBDImage(tmpRgb, tmpDepth, tmpModel);
if(info) info->timeMirroring = timer.ticks();
}
@@ -280,12 +286,11 @@ void CameraThread::postUpdate(SensorData * dataPtr, CameraInfo * info) const
images.push_back(data.imageRaw().getUMat(cv::ACCESS_READ));
images.push_back(data.rightRaw().getUMat(cv::ACCESS_READ));
compensator->feed(topLeftCorners, images, masks);
cv::Mat img = data.imageRaw().clone();
compensator->apply(0, cv::Point(0,0), img, masks[0]);
data.setImageRaw(img);
img = data.rightRaw().clone();
compensator->apply(1, cv::Point(0,0), img, masks[1]);
data.setDepthOrRightRaw(img);
cv::Mat imgLeft = data.imageRaw().clone();
compensator->apply(0, cv::Point(0,0), imgLeft, masks[0]);
cv::Mat imgRight = data.rightRaw().clone();
compensator->apply(1, cv::Point(0,0), imgRight, masks[1]);
data.setStereoImage(imgLeft, imgRight, data.stereoCameraModel());
cv::detail::GainCompensator * gainCompensator = (cv::detail::GainCompensator*)compensator.get();
UDEBUG("gains = %f %f ", gainCompensator->gains()[0], gainCompensator->gains()[1]);
if(info) info->timeStereoExposureCompensation = timer.ticks();
@@ -309,9 +314,7 @@ void CameraThread::postUpdate(SensorData * dataPtr, CameraInfo * info) const
data.stereoCameraModel().localTransform(),
-data.stereoCameraModel().baseline()*data.stereoCameraModel().left().fx(),
data.stereoCameraModel().left().imageSize());
data.setCameraModel(model);
data.setDepthOrRightRaw(depth);
data.setStereoCameraModel(StereoCameraModel());
data.setRGBDImage(data.imageRaw(), depth, model);
if(info) info->timeDisparity = timer.ticks();
}
if(_scanFromDepth &&
@@ -367,7 +370,7 @@ void CameraThread::postUpdate(SensorData * dataPtr, CameraInfo * info) const
}
}
}
data.setLaserScanRaw(LaserScan(scan, (int)maxPoints, _scanRangeMax, format, baseToScan));
data.setLaserScan(LaserScan(scan, (int)maxPoints, _scanRangeMax, format, baseToScan));
if(info) info->timeScanFromDepth = timer.ticks();
}
else
@@ -381,7 +384,7 @@ void CameraThread::postUpdate(SensorData * dataPtr, CameraInfo * info) const
{
UDEBUG("");
// filter the scan after registration
data.setLaserScanRaw(util3d::commonFiltering(data.laserScanRaw(), _scanDownsampleStep, _scanRangeMin, _scanRangeMax, _scanVoxelSize, _scanNormalsK, _scanNormalsRadius, _scanForceGroundNormalsUp));
data.setLaserScan(util3d::commonFiltering(data.laserScanRaw(), _scanDownsampleStep, _scanRangeMin, _scanRangeMax, _scanVoxelSize, _scanNormalsK, _scanNormalsRadius, _scanForceGroundNormalsUp));
}
}

39
corelib/src/DBDriverSqlite3.cpp
View File

@@ -1760,10 +1760,9 @@ void DBDriverSqlite3::loadNodeDataQuery(std::list<Signature *> & signatures, boo
viewPoint.z = sqlite3_column_double(ppStmt, index++);
}
SensorData tmp = (*iter)->sensorData();
LaserScan laserScan = tmp.laserScanCompressed();
if(scan)
{
LaserScan laserScan;
if(laserScanAngleMin < laserScanAngleMax && laserScanAngleInc != 0.0f)
{
laserScan = LaserScan(scanCompressed, (LaserScan::Format)laserScanFormat, laserScanMinRange, laserScanMaxRange, laserScanAngleMin, laserScanAngleMax, laserScanAngleInc, scanLocalTransform);
@@ -1772,37 +1771,29 @@ void DBDriverSqlite3::loadNodeDataQuery(std::list<Signature *> & signatures, boo
{
laserScan = LaserScan(scanCompressed, laserScanMaxPts, laserScanMaxRange, (LaserScan::Format)laserScanFormat, scanLocalTransform);
}
(*iter)->sensorData().setLaserScan(laserScan);
}
if(models.size())
if(images)
{
(*iter)->sensorData() = SensorData(
laserScan,
images?imageCompressed:tmp.imageCompressed(),
images?depthOrRightCompressed:tmp.depthOrRightCompressed(),
images?models:tmp.cameraModels(),
(*iter)->id(),
(*iter)->getStamp(),
userData?userDataCompressed:tmp.userDataCompressed());
if(models.size())
{
(*iter)->sensorData().setRGBDImage(imageCompressed, depthOrRightCompressed, models);
}
else
{
(*iter)->sensorData().setStereoImage(imageCompressed, depthOrRightCompressed, stereoModel);
}
}
else
if(userData)
{
(*iter)->sensorData() = SensorData(
laserScan,
images?imageCompressed:tmp.imageCompressed(),
images?depthOrRightCompressed:tmp.depthOrRightCompressed(),
images?stereoModel:tmp.stereoCameraModel(),
(*iter)->id(),
(*iter)->getStamp(),
userData?userDataCompressed:tmp.userDataCompressed());
(*iter)->sensorData().setUserData(userDataCompressed);
}
if(occupancyGrid)
{
(*iter)->sensorData().setOccupancyGrid(groundCellsCompressed, obstacleCellsCompressed, emptyCellsCompressed, cellSize, viewPoint);
}
else
{
(*iter)->sensorData().setOccupancyGrid(tmp.gridGroundCellsCompressed(), tmp.gridObstacleCellsCompressed(), tmp.gridEmptyCellsCompressed(), tmp.gridCellSize(), tmp.gridViewPoint());
}
rc = sqlite3_step(ppStmt); // next result...
}
UASSERT_MSG(rc == SQLITE_DONE, uFormat("DB error (%s): %s", _version.c_str(), sqlite3_errmsg(_ppDb)).c_str());

15
corelib/src/DBReader.cpp
View File

@@ -423,24 +423,23 @@ SensorData DBReader::getNextData(CameraInfo * info)
{
// select one camera
int subImageWidth = data.imageRaw().cols/data.cameraModels().size();
cv::Mat image;
UASSERT(!data.imageRaw().empty() &&
data.imageRaw().cols % data.cameraModels().size() == 0 &&
_cameraIndex*subImageWidth < data.imageRaw().cols);
data.setImageRaw(
cv::Mat(data.imageRaw(),
cv::Rect(_cameraIndex*subImageWidth, 0, subImageWidth, data.imageRaw().rows)).clone());
image= cv::Mat(data.imageRaw(),
cv::Rect(_cameraIndex*subImageWidth, 0, subImageWidth, data.imageRaw().rows)).clone();
cv::Mat depth;
if(!data.depthOrRightRaw().empty())
{
UASSERT(data.depthOrRightRaw().cols % data.cameraModels().size() == 0 &&
subImageWidth == data.depthOrRightRaw().cols/(int)data.cameraModels().size() &&
_cameraIndex*subImageWidth < data.depthOrRightRaw().cols);
data.setDepthOrRightRaw(
cv::Mat(data.depthOrRightRaw(),
cv::Rect(_cameraIndex*subImageWidth, 0, subImageWidth, data.depthOrRightRaw().rows)).clone());
depth = cv::Mat(data.depthOrRightRaw(),
cv::Rect(_cameraIndex*subImageWidth, 0, subImageWidth, data.depthOrRightRaw().rows)).clone();
}
CameraModel model = data.cameraModels().at(_cameraIndex);
data.setCameraModel(model);
data.setRGBDImage(image, depth, data.cameraModels().at(_cameraIndex));
}
else
{

90
corelib/src/Memory.cpp
View File

@@ -2643,21 +2643,10 @@ void Memory::removeRawData(int id, bool image, bool scan, bool userData)
Signature * s = this->_getSignature(id);
if(s)
{
if(image && (!_reextractLoopClosureFeatures || !_registrationPipeline->isImageRequired()))
{
s->sensorData().setImageRaw(cv::Mat());
s->sensorData().setDepthOrRightRaw(cv::Mat());
}
if(scan && !_registrationPipeline->isScanRequired())
{
LaserScan scan = s->sensorData().laserScanRaw();
scan.clear();
s->sensorData().setLaserScanRaw(scan);
}
if(userData && !_registrationPipeline->isUserDataRequired())
{
s->sensorData().setUserDataRaw(cv::Mat());
}
s->sensorData().clearRawData(
image && (!_reextractLoopClosureFeatures || !_registrationPipeline->isImageRequired()),
scan && !_registrationPipeline->isScanRequired(),
userData && !_registrationPipeline->isUserDataRequired());
}
}
@@ -3135,7 +3124,7 @@ Transform Memory::computeIcpTransformMulti(
}
// scans are in base frame but for 2d scans, set the height so that correspondences matching works
assembledData.setLaserScanRaw(
assembledData.setLaserScan(
LaserScan(assembledScan,
fromScan.maxPoints()?fromScan.maxPoints():maxPoints,
fromScan.rangeMax(),
@@ -4103,7 +4092,7 @@ Signature * Memory::createSignature(const SensorData & inputData, const Transfor
return 0;
}
}
data.setImageRaw(rectifiedImages);
data.setRGBDImage(rectifiedImages, data.depthOrRightRaw(), data.cameraModels());
}
else if(data.stereoCameraModel().isValidForRectification())
{
@@ -4119,8 +4108,10 @@ Signature * Memory::createSignature(const SensorData & inputData, const Transfor
}
UASSERT(_rectStereoCameraModel.left().imageWidth() == data.stereoCameraModel().left().imageWidth());
UASSERT(_rectStereoCameraModel.left().imageHeight() == data.stereoCameraModel().left().imageHeight());
data.setImageRaw(_rectStereoCameraModel.left().rectifyImage(data.imageRaw()));
data.setDepthOrRightRaw(_rectStereoCameraModel.right().rectifyImage(data.rightRaw()));
data.setStereoImage(
_rectStereoCameraModel.left().rectifyImage(data.imageRaw()),
_rectStereoCameraModel.right().rectifyImage(data.rightRaw()),
data.stereoCameraModel());
imagesRectified = true;
}
else
@@ -4159,24 +4150,41 @@ Signature * Memory::createSignature(const SensorData & inputData, const Transfor
if(_imagePreDecimation > 1)
{
preDecimation = _imagePreDecimation;
if(!decimatedData.rightRaw().empty() ||
(decimatedData.depthRaw().rows == decimatedData.imageRaw().rows && decimatedData.depthRaw().cols == decimatedData.imageRaw().cols))
{
decimatedData.setDepthOrRightRaw(util2d::decimate(decimatedData.depthOrRightRaw(), _imagePreDecimation));
}
decimatedData.setImageRaw(util2d::decimate(decimatedData.imageRaw(), _imagePreDecimation));
std::vector<CameraModel> cameraModels = decimatedData.cameraModels();
for(unsigned int i=0; i<cameraModels.size(); ++i)
{
cameraModels[i] = cameraModels[i].scaled(1.0/double(_imagePreDecimation));
}
decimatedData.setCameraModels(cameraModels);
StereoCameraModel stereoModel = decimatedData.stereoCameraModel();
if(stereoModel.isValidForProjection())
if(!cameraModels.empty())
{
stereoModel.scale(1.0/double(_imagePreDecimation));
if(decimatedData.depthRaw().rows == decimatedData.imageRaw().rows &&
decimatedData.depthRaw().cols == decimatedData.imageRaw().cols)
{
decimatedData.setRGBDImage(
util2d::decimate(decimatedData.imageRaw(), _imagePreDecimation),
util2d::decimate(decimatedData.depthOrRightRaw(), _imagePreDecimation),
cameraModels);
}
else
{
decimatedData.setRGBDImage(
util2d::decimate(decimatedData.imageRaw(), _imagePreDecimation),
decimatedData.depthOrRightRaw(),
cameraModels);
}
}
else
{
StereoCameraModel stereoModel = decimatedData.stereoCameraModel();
if(stereoModel.isValidForProjection())
{
stereoModel.scale(1.0/double(_imagePreDecimation));
}
decimatedData.setStereoImage(
util2d::decimate(decimatedData.imageRaw(), _imagePreDecimation),
util2d::decimate(decimatedData.depthOrRightRaw(), _imagePreDecimation),
stereoModel);
}
decimatedData.setStereoCameraModel(stereoModel);
}
UINFO("Extract features");
@@ -5009,10 +5017,16 @@ Signature * Memory::createSignature(const SensorData & inputData, const Transfor
s->setWordsDescriptors(wordsDescriptors);
// set raw data
s->sensorData().setImageRaw(image);
s->sensorData().setDepthOrRightRaw(depthOrRightImage);
s->sensorData().setLaserScanRaw(laserScan);
s->sensorData().setUserDataRaw(data.userDataRaw());
if(!cameraModels.empty())
{
s->sensorData().setRGBDImage(image, depthOrRightImage, cameraModels, false);
}
else
{
s->sensorData().setStereoImage(image, depthOrRightImage, stereoCameraModel, false);
}
s->sensorData().setLaserScan(laserScan, false);
s->sensorData().setUserData(data.userDataRaw(), false);
s->sensorData().setGroundTruth(data.groundTruth());
s->sensorData().setGPS(data.gps());
@@ -5059,6 +5073,7 @@ Signature * Memory::createSignature(const SensorData & inputData, const Transfor
if(!data.globalPose().isNull() && data.globalPoseCovariance().cols==6 && data.globalPoseCovariance().rows==6 && data.globalPoseCovariance().cols==CV_64FC1)
{
s->addLink(Link(s->id(), s->id(), Link::kPosePrior, data.globalPose(), data.globalPoseCovariance().inv()));
UDEBUG("Added global pose prior: %s", data.globalPose().prettyPrint().c_str());
if(data.gps().stamp() > 0.0)
{
@@ -5070,14 +5085,17 @@ Signature * Memory::createSignature(const SensorData & inputData, const Transfor
if(_gpsOrigin.stamp() <= 0.0)
{
_gpsOrigin = data.gps();
UINFO("Added GPS origin: long=%f lat=%f alt=%f bearing=%f error=%f", data.gps().longitude(), data.gps().latitude(), data.gps().altitude(), data.gps().bearing(), data.gps().error());
}
cv::Point3f pt = data.gps().toGeodeticCoords().toENU_WGS84(_gpsOrigin.toGeodeticCoords());
Transform gpsPose(pt.x, pt.y, pose.z(), 0, 0, -(data.gps().bearing()-90.0)*180.0/M_PI);
cv::Mat gpsInfMatrix = cv::Mat::eye(6,6,CV_64FC1)/9999.0; // variance not used >= 9999
if(data.gps().error() > 0.0)
{
// only set x, y and z as we don't know variance for other degrees of freedom.
gpsInfMatrix.at<double>(0,0) = gpsInfMatrix.at<double>(1,1) = gpsInfMatrix.at<double>(2,2) = 1.0/data.gps().error();
UDEBUG("Added GPS prior: x=%f y=%f z=%f yaw=%f", gpsPose.x(), gpsPose.y(), gpsPose.z(), gpsPose.theta());
// only set x, y as we don't know variance for other degrees of freedom.
gpsInfMatrix.at<double>(0,0) = gpsInfMatrix.at<double>(1,1) = 1.0/data.gps().error();
gpsInfMatrix.at<double>(2,2) = 1; // z variance is set to avoid issues with g2o and gtsam requiring a prior on Z
s->addLink(Link(s->id(), s->id(), Link::kPosePrior, gpsPose, gpsInfMatrix));
}
else

21
corelib/src/Odometry.cpp
View File

@@ -405,20 +405,27 @@ Transform Odometry::process(SensorData & data, const Transform & guessIn, Odomet
{
// Decimation of images with calibrations
SensorData decimatedData = data;
decimatedData.setImageRaw(util2d::decimate(decimatedData.imageRaw(), _imageDecimation));
decimatedData.setDepthOrRightRaw(util2d::decimate(decimatedData.depthOrRightRaw(), _imageDecimation));
cv::Mat rgbLeft = util2d::decimate(decimatedData.imageRaw(), _imageDecimation);
cv::Mat depthRight = util2d::decimate(decimatedData.depthOrRightRaw(), _imageDecimation);
std::vector<CameraModel> cameraModels = decimatedData.cameraModels();
for(unsigned int i=0; i<cameraModels.size(); ++i)
{
cameraModels[i] = cameraModels[i].scaled(1.0/double(_imageDecimation));
}
decimatedData.setCameraModels(cameraModels);
StereoCameraModel stereoModel = decimatedData.stereoCameraModel();
if(stereoModel.isValidForProjection())
if(!cameraModels.empty())
{
stereoModel.scale(1.0/double(_imageDecimation));
decimatedData.setRGBDImage(rgbLeft, depthRight, cameraModels);
}
decimatedData.setStereoCameraModel(stereoModel);
else
{
StereoCameraModel stereoModel = decimatedData.stereoCameraModel();
if(stereoModel.isValidForProjection())
{
stereoModel.scale(1.0/double(_imageDecimation));
}
decimatedData.setStereoImage(rgbLeft, depthRight, stereoModel);
}
// compute transform
t = this->computeTransform(decimatedData, guess, info);

30
corelib/src/Optimizer.cpp
View File

@@ -176,13 +176,16 @@ void Optimizer::getConnectedGraph(
std::multimap<int, int> biLinks;
for(std::multimap<int, Link>::const_iterator iter=linksIn.begin(); iter!=linksIn.end(); ++iter)
{
UASSERT_MSG(graph::findLink(biLinks, iter->second.from(), iter->second.to()) == biLinks.end(),
uFormat("Input links should be unique between two poses (%d->%d).",
iter->second.from(), iter->second.to()).c_str());
biLinks.insert(std::make_pair(iter->second.from(), iter->second.to()));
if(iter->second.from() != iter->second.to())
{
biLinks.insert(std::make_pair(iter->second.to(), iter->second.from()));
UASSERT_MSG(graph::findLink(biLinks, iter->second.from(), iter->second.to()) == biLinks.end(),
uFormat("Input links should be unique between two poses (%d->%d).",
iter->second.from(), iter->second.to()).c_str());
biLinks.insert(std::make_pair(iter->second.from(), iter->second.to()));
if(iter->second.from() != iter->second.to())
{
biLinks.insert(std::make_pair(iter->second.to(), iter->second.from()));
}
}
}
@@ -197,6 +200,14 @@ void Optimizer::getConnectedGraph(
if(posesOut.empty())
{
posesOut.insert(*posesIn.find(fromId));
// add prior links
for(std::multimap<int, Link>::const_iterator pter=linksIn.find(fromId); pter!=linksIn.end() && pter->first==fromId; ++pter)
{
if(pter->second.from() == pter->second.to())
{
linksOut.insert(*pter);
}
}
}
for(std::multimap<int, int>::const_iterator iter=biLinks.find(fromId); iter!=biLinks.end() && iter->first==fromId; ++iter)
@@ -211,6 +222,15 @@ void Optimizer::getConnectedGraph(
if(!uContains(posesOut, toId))
{
posesOut.insert(*posesIn.find(toId));//std::make_pair(toId, posesOut.at(fromId) * (kter->second.from()==fromId?kter->second.transform():kter->second.transform().inverse())));
// add prior links
for(std::multimap<int, Link>::const_iterator pter=linksIn.find(toId); pter!=linksIn.end() && pter->first==toId; ++pter)
{
if(pter->second.from() == pter->second.to())
{
linksOut.insert(*pter);
}
}
if(curentPoses.find(toId) == curentPoses.end())
{
nextPoses.insert(toId);

16
corelib/src/RegistrationIcp.cpp
View File

@@ -794,7 +794,7 @@ Transform RegistrationIcp::computeTransformationImpl(
// update output scans
if(fromScan.is2d())
{
fromSignature.sensorData().setLaserScanRaw(
fromSignature.sensorData().setLaserScan(
LaserScan(
util3d::laserScan2dFromPointCloud(*fromCloudNormals, fromScan.localTransform().inverse()),
maxLaserScansFrom,
@@ -804,7 +804,7 @@ Transform RegistrationIcp::computeTransformationImpl(
}
else
{
fromSignature.sensorData().setLaserScanRaw(
fromSignature.sensorData().setLaserScan(
LaserScan(
util3d::laserScanFromPointCloud(*fromCloudNormals, fromScan.localTransform().inverse()),
maxLaserScansFrom,
@@ -814,7 +814,7 @@ Transform RegistrationIcp::computeTransformationImpl(
}
if(toScan.is2d())
{
toSignature.sensorData().setLaserScanRaw(
toSignature.sensorData().setLaserScan(
LaserScan(
util3d::laserScan2dFromPointCloud(*toCloudNormals, (guess*toScan.localTransform()).inverse()),
maxLaserScansTo,
@@ -824,7 +824,7 @@ Transform RegistrationIcp::computeTransformationImpl(
}
else
{
toSignature.sensorData().setLaserScanRaw(
toSignature.sensorData().setLaserScan(
LaserScan(
util3d::laserScanFromPointCloud(*toCloudNormals, (guess*toScan.localTransform()).inverse()),
maxLaserScansTo,
@@ -913,7 +913,7 @@ Transform RegistrationIcp::computeTransformationImpl(
// update output scans
if(fromScan.is2d())
{
fromSignature.sensorData().setLaserScanRaw(
fromSignature.sensorData().setLaserScan(
LaserScan(
util3d::laserScan2dFromPointCloud(*fromCloudFiltered, fromScan.localTransform().inverse()),
maxLaserScansFrom,
@@ -923,7 +923,7 @@ Transform RegistrationIcp::computeTransformationImpl(
}
else
{
fromSignature.sensorData().setLaserScanRaw(
fromSignature.sensorData().setLaserScan(
LaserScan(
util3d::laserScanFromPointCloud(*fromCloudFiltered, fromScan.localTransform().inverse()),
maxLaserScansFrom,
@@ -933,7 +933,7 @@ Transform RegistrationIcp::computeTransformationImpl(
}
if(toScan.is2d())
{
toSignature.sensorData().setLaserScanRaw(
toSignature.sensorData().setLaserScan(
LaserScan(
util3d::laserScan2dFromPointCloud(*toCloudFiltered, (guess*toScan.localTransform()).inverse()),
maxLaserScansTo,
@@ -943,7 +943,7 @@ Transform RegistrationIcp::computeTransformationImpl(
}
else
{
toSignature.sensorData().setLaserScanRaw(
toSignature.sensorData().setLaserScan(
LaserScan(
util3d::laserScanFromPointCloud(*toCloudFiltered, (guess*toScan.localTransform()).inverse()),
maxLaserScansTo,

542
corelib/src/SensorData.cpp
View File

@@ -54,26 +54,8 @@ SensorData::SensorData(
_stamp(stamp),
_cellSize(0.0f)
{
if(image.rows == 1)
{
UASSERT(image.type() == CV_8UC1); // Bytes
_imageCompressed = image;
}
else if(!image.empty())
{
UASSERT(image.type() == CV_8UC1 || // Mono
image.type() == CV_8UC3); // RGB
_imageRaw = image;
}
if(userData.type() == CV_8UC1 && userData.rows == 1 && userData.cols > int(3*sizeof(int))) // Bytes
{
_userDataCompressed = userData; // assume compressed
}
else
{
_userDataRaw = userData;
}
setRGBDImage(image, cv::Mat(), CameraModel());
setUserData(userData);
}
// Mono constructor
@@ -85,29 +67,10 @@ SensorData::SensorData(
const cv::Mat & userData) :
_id(id),
_stamp(stamp),
_cameraModels(std::vector<CameraModel>(1, cameraModel)),
_cellSize(0.0f)
{
if(image.rows == 1)
{
UASSERT(image.type() == CV_8UC1); // Bytes
_imageCompressed = image;
}
else if(!image.empty())
{
UASSERT(image.type() == CV_8UC1 || // Mono
image.type() == CV_8UC3); // RGB
_imageRaw = image;
}
if(userData.type() == CV_8UC1 && userData.rows == 1 && userData.cols > int(3*sizeof(int))) // Bytes
{
_userDataCompressed = userData; // assume compressed
}
else
{
_userDataRaw = userData;
}
setRGBDImage(image, cv::Mat(), cameraModel);
setUserData(userData);
}
// RGB-D constructor
@@ -120,41 +83,10 @@ SensorData::SensorData(
const cv::Mat & userData) :
_id(id),
_stamp(stamp),
_cameraModels(std::vector<CameraModel>(1, cameraModel)),
_cellSize(0.0f)
{
if(rgb.rows == 1)
{
UASSERT(rgb.type() == CV_8UC1); // Bytes
_imageCompressed = rgb;
}
else if(!rgb.empty())
{
UASSERT(rgb.type() == CV_8UC1 || // Mono
rgb.type() == CV_8UC3); // RGB
_imageRaw = rgb;
}
if(depth.rows == 1)
{
UASSERT(depth.type() == CV_8UC1); // Bytes
_depthOrRightCompressed = depth;
}
else if(!depth.empty())
{
UASSERT(depth.type() == CV_32FC1 || // Depth in meter
depth.type() == CV_16UC1); // Depth in millimetre
_depthOrRightRaw = depth;
}
if(userData.type() == CV_8UC1 && userData.rows == 1 && userData.cols > int(3*sizeof(int))) // Bytes
{
_userDataCompressed = userData; // assume compressed
}
else
{
_userDataRaw = userData;
}
setRGBDImage(rgb, depth, cameraModel);
setUserData(userData);
}
// RGB-D constructor + laser scan
@@ -168,49 +100,11 @@ SensorData::SensorData(
const cv::Mat & userData) :
_id(id),
_stamp(stamp),
_cameraModels(std::vector<CameraModel>(1, cameraModel)),
_cellSize(0.0f)
{
if(rgb.rows == 1)
{
UASSERT(rgb.type() == CV_8UC1); // Bytes
_imageCompressed = rgb;
}
else if(!rgb.empty())
{
UASSERT(rgb.type() == CV_8UC1 || // Mono
rgb.type() == CV_8UC3); // RGB
_imageRaw = rgb;
}
if(depth.rows == 1)
{
UASSERT(depth.type() == CV_8UC1); // Bytes
_depthOrRightCompressed = depth;
}
else if(!depth.empty())
{
UASSERT(depth.type() == CV_32FC1 || // Depth in meter
depth.type() == CV_16UC1); // Depth in millimetre
_depthOrRightRaw = depth;
}
if(!laserScan.isCompressed())
{
_laserScanRaw = laserScan;
}
else
{
_laserScanCompressed = laserScan;
}
if(userData.type() == CV_8UC1 && userData.rows == 1 && userData.cols > int(3*sizeof(int))) // Bytes
{
_userDataCompressed = userData; // assume compressed
}
else
{
_userDataRaw = userData;
}
setRGBDImage(rgb, depth, cameraModel);
setLaserScan(laserScan);
setUserData(userData);
}
// Multi-cameras RGB-D constructor
@@ -223,40 +117,10 @@ SensorData::SensorData(
const cv::Mat & userData) :
_id(id),
_stamp(stamp),
_cameraModels(cameraModels),
_cellSize(0.0f)
{
if(rgb.rows == 1)
{
UASSERT(rgb.type() == CV_8UC1); // Bytes
_imageCompressed = rgb;
}
else if(!rgb.empty())
{
UASSERT(rgb.type() == CV_8UC1 || // Mono
rgb.type() == CV_8UC3); // RGB
_imageRaw = rgb;
}
if(depth.rows == 1)
{
UASSERT(depth.type() == CV_8UC1); // Bytes
_depthOrRightCompressed = depth;
}
else if(!depth.empty())
{
UASSERT(depth.type() == CV_32FC1 || // Depth in meter
depth.type() == CV_16UC1); // Depth in millimetre
_depthOrRightRaw = depth;
}
if(userData.type() == CV_8UC1 && userData.rows == 1 && userData.cols > int(3*sizeof(int))) // Bytes
{
_userDataCompressed = userData; // assume compressed
}
else
{
_userDataRaw = userData;
}
setRGBDImage(rgb, depth, cameraModels);
setUserData(userData);
}
// Multi-cameras RGB-D constructor + laser scan
@@ -270,49 +134,11 @@ SensorData::SensorData(
const cv::Mat & userData) :
_id(id),
_stamp(stamp),
_cameraModels(cameraModels),
_cellSize(0.0f)
{
if(rgb.rows == 1)
{
UASSERT(rgb.type() == CV_8UC1); // Bytes
_imageCompressed = rgb;
}
else if(!rgb.empty())
{
UASSERT(rgb.type() == CV_8UC1 || // Mono
rgb.type() == CV_8UC3); // RGB
_imageRaw = rgb;
}
if(depth.rows == 1)
{
UASSERT(depth.type() == CV_8UC1); // Bytes
_depthOrRightCompressed = depth;
}
else if(!depth.empty())
{
UASSERT(depth.type() == CV_32FC1 || // Depth in meter
depth.type() == CV_16UC1); // Depth in millimetre
_depthOrRightRaw = depth;
}
if(!laserScan.isCompressed())
{
_laserScanRaw = laserScan;
}
else
{
_laserScanCompressed = laserScan;
}
if(userData.type() == CV_8UC1 && userData.rows == 1 && userData.cols > int(3*sizeof(int))) // Bytes
{
_userDataCompressed = userData; // assume compressed
}
else
{
_userDataRaw = userData;
}
setRGBDImage(rgb, depth, cameraModels);
setLaserScan(laserScan);
setUserData(userData);
}
// Stereo constructor
@@ -325,42 +151,10 @@ SensorData::SensorData(
const cv::Mat & userData):
_id(id),
_stamp(stamp),
_stereoCameraModel(cameraModel),
_cellSize(0.0f)
{
if(left.rows == 1)
{
UASSERT(left.type() == CV_8UC1); // Bytes
_imageCompressed = left;
}
else if(!left.empty())
{
UASSERT(left.type() == CV_8UC1 || // Mono
left.type() == CV_8UC3 || // RGB
left.type() == CV_16UC1); // IR
_imageRaw = left;
}
if(right.rows == 1)
{
UASSERT(right.type() == CV_8UC1); // Bytes
_depthOrRightCompressed = right;
}
else if(!right.empty())
{
UASSERT(right.type() == CV_8UC1 || // Mono
right.type() == CV_16UC1); // IR
_depthOrRightRaw = right;
}
if(userData.type() == CV_8UC1 && userData.rows == 1 && userData.cols > int(3*sizeof(int))) // Bytes
{
_userDataCompressed = userData; // assume compressed
}
else
{
_userDataRaw = userData;
}
setStereoImage(left, right, cameraModel);
setUserData(userData);
}
// Stereo constructor + 2d laser scan
@@ -374,48 +168,11 @@ SensorData::SensorData(
const cv::Mat & userData) :
_id(id),
_stamp(stamp),
_stereoCameraModel(cameraModel),
_cellSize(0.0f)
{
if(left.rows == 1)
{
UASSERT(left.type() == CV_8UC1); // Bytes
_imageCompressed = left;
}
else if(!left.empty())
{
UASSERT(left.type() == CV_8UC1 || // Mono
left.type() == CV_8UC3); // RGB
_imageRaw = left;
}
if(right.rows == 1)
{
UASSERT(right.type() == CV_8UC1); // Bytes
_depthOrRightCompressed = right;
}
else if(!right.empty())
{
UASSERT(right.type() == CV_8UC1); // Mono
_depthOrRightRaw = right;
}
if(!laserScan.isCompressed())
{
_laserScanRaw = laserScan;
}
else
{
_laserScanCompressed = laserScan;
}
if(userData.type() == CV_8UC1 && userData.rows == 1 && userData.cols > int(3*sizeof(int))) // Bytes
{
_userDataCompressed = userData; // assume compressed
}
else
{
_userDataRaw = userData;
}
setStereoImage(left, right, cameraModel);
setLaserScan(laserScan);
setUserData(userData);
}
SensorData::SensorData(
@@ -433,47 +190,208 @@ SensorData::~SensorData()
{
}
void SensorData::setRGBDImage(
const cv::Mat & rgb,
const cv::Mat & depth,
const CameraModel & model,
bool clearPreviousData)
{
std::vector<CameraModel> models;
models.push_back(model);
setRGBDImage(rgb, depth, models, clearPreviousData);
}
void SensorData::setRGBDImage(
const cv::Mat & rgb,
const cv::Mat & depth,
const std::vector<CameraModel> & models,
bool clearPreviousData)
{
if(!clearPreviousData && _stereoCameraModel.isValidForProjection())
{
UERROR("Sensor data has previously stereo images "
"but clearPreviousData parameter is false. We "
"will still clear previous data to avoid incompatibilities "
"between raw and compressed data!");
}
bool clearData = clearPreviousData || _stereoCameraModel.isValidForProjection();
_stereoCameraModel = StereoCameraModel();
_cameraModels = models;
if(rgb.rows == 1)
{
UASSERT(rgb.type() == CV_8UC1); // Bytes
_imageCompressed = rgb;
if(clearData)
{
_imageRaw = cv::Mat();
}
}
else if(!rgb.empty())
{
UASSERT(rgb.type() == CV_8UC1 || // Mono
rgb.type() == CV_8UC3); // RGB
_imageRaw = rgb;
if(clearData)
{
_imageCompressed = cv::Mat();
}
}
else if(clearData)
{
_imageRaw = cv::Mat();
_imageCompressed = cv::Mat();
}
if(depth.rows == 1)
{
UASSERT(depth.type() == CV_8UC1); // Bytes
_depthOrRightCompressed = depth;
if(clearData)
{
_depthOrRightRaw = cv::Mat();
}
}
else if(!depth.empty())
{
UASSERT(depth.type() == CV_32FC1 || // Depth in meter
depth.type() == CV_16UC1); // Depth in millimetre
_depthOrRightRaw = depth;
if(clearData)
{
_depthOrRightCompressed = cv::Mat();
}
}
else if(clearData)
{
_depthOrRightRaw = cv::Mat();
_depthOrRightCompressed = cv::Mat();
}
}
void SensorData::setStereoImage(
const cv::Mat & left,
const cv::Mat & right,
const StereoCameraModel & stereoCameraModel,
bool clearPreviousData)
{
if(!clearPreviousData && !_cameraModels.empty())
{
UERROR("Sensor data has previously RGB-D/RGB images "
"but clearPreviousData parameter is false. We "
"will still clear previous data to avoid incompatibilities "
"between raw and compressed data!");
}
bool clearData = clearPreviousData || !_cameraModels.empty();
_cameraModels.clear();
_stereoCameraModel = stereoCameraModel;
if(left.rows == 1)
{
UASSERT(left.type() == CV_8UC1); // Bytes
_imageCompressed = left;
if(clearData)
{
_imageRaw = cv::Mat();
}
}
else if(!left.empty())
{
UASSERT(left.type() == CV_8UC1 || // Mono
left.type() == CV_8UC3); // RGB
_imageRaw = left;
if(clearData)
{
_imageCompressed = cv::Mat();
}
}
else if(clearData)
{
_imageRaw = cv::Mat();
_imageCompressed = cv::Mat();
}
if(right.rows == 1)
{
UASSERT(right.type() == CV_8UC1); // Bytes
_depthOrRightCompressed = right;
if(clearData)
{
_depthOrRightRaw = cv::Mat();
}
}
else if(!right.empty())
{
UASSERT(right.type() == CV_8UC1); // Mono
_depthOrRightRaw = right;
if(clearData)
{
_depthOrRightCompressed = cv::Mat();
}
}
else if(clearData)
{
_depthOrRightRaw = cv::Mat();
_depthOrRightCompressed = cv::Mat();
}
}
void SensorData::setLaserScan(const LaserScan & laserScan, bool clearPreviousData)
{
if(!laserScan.isCompressed())
{
_laserScanRaw = laserScan;
if(clearPreviousData)
{
_laserScanCompressed = LaserScan();
}
}
else
{
_laserScanCompressed = laserScan;
if(clearPreviousData)
{
_laserScanRaw = LaserScan();
}
}
}
void SensorData::setImageRaw(const cv::Mat & image)
{
UASSERT(image.empty() || image.rows > 1);
_imageRaw = image;
}
void SensorData::setDepthOrRightRaw(const cv::Mat & image)
{
UASSERT(image.empty() || image.rows > 1);
_depthOrRightRaw = image;
}
void SensorData::setLaserScanRaw(const LaserScan & scan)
{
UASSERT(scan.isEmpty() || !scan.isCompressed());
_laserScanRaw = scan;
}
void SensorData::setUserDataRaw(const cv::Mat & userDataRaw)
{
if(!userDataRaw.empty() && !_userDataRaw.empty())
{
UWARN("Cannot write new user data (%d bytes) over existing user "
"data (%d bytes, %d compressed). Set user data of %d to null "
"before setting a new one.",
int(userDataRaw.total()*userDataRaw.elemSize()),
int(_userDataRaw.total()*_userDataRaw.elemSize()),
_userDataCompressed.cols,
this->id());
return;
}
_userDataRaw = userDataRaw;
}
void SensorData::setUserData(const cv::Mat & userData)
void SensorData::setUserData(const cv::Mat & userData, bool clearPreviousData)
{
if(!userData.empty() && (!_userDataCompressed.empty() || !_userDataRaw.empty()))
if(clearPreviousData)
{
UWARN("Cannot write new user data (%d bytes) over existing user "
"data (%d bytes, %d compressed). Set user data of %d to null "
"before setting a new one.",
int(userData.total()*userData.elemSize()),
int(_userDataRaw.total()*_userDataRaw.elemSize()),
_userDataCompressed.cols,
this->id());
return;
_userDataRaw = cv::Mat();
_userDataCompressed = cv::Mat();
}
_userDataRaw = cv::Mat();
_userDataCompressed = cv::Mat();
if(!userData.empty())
if(userData.type() == CV_8UC1 && userData.rows == 1 && userData.cols > int(3*sizeof(int))) // Bytes
{
if(userData.type() == CV_8UC1 && userData.rows == 1 && userData.cols > int(3*sizeof(int))) // Bytes
_userDataCompressed = userData; // assume compressed
}
else
{
_userDataRaw = userData;
if(!userData.empty())
{
_userDataCompressed = userData; // assume compressed
}
else
{
_userDataRaw = userData;
_userDataCompressed = compressData2(userData);
}
}
@@ -491,7 +409,8 @@ void SensorData::setOccupancyGrid(
(!obstacles.empty() && (!_obstacleCellsCompressed.empty() || !_obstacleCellsRaw.empty())) ||
(!empty.empty() && (!_emptyCellsCompressed.empty() || !_emptyCellsRaw.empty())))
{
UWARN("Occupancy grid cannot be overwritten! id=%d", this->id());
UWARN("Occupancy grid cannot be overwritten! id=%d, Set occupancy grid of %d to null "
"before setting a new one.", this->id());
return;
}
@@ -515,7 +434,6 @@ void SensorData::setOccupancyGrid(
}
else if(ground.type() == CV_8UC1)
{
UASSERT(ground.type() == CV_8UC1); // Bytes
_groundCellsCompressed = ground;
}
}
@@ -528,7 +446,6 @@ void SensorData::setOccupancyGrid(
}
else if(obstacles.type() == CV_8UC1)
{
UASSERT(obstacles.type() == CV_8UC1); // Bytes
_obstacleCellsCompressed = obstacles;
}
}
@@ -541,7 +458,6 @@ void SensorData::setOccupancyGrid(
}
else if(empty.type() == CV_8UC1)
{
UASSERT(empty.type() == CV_8UC1); // Bytes
_emptyCellsCompressed = empty;
}
}
@@ -868,6 +784,40 @@ long SensorData::getMemoryUsed() const // Return memory usage in Bytes
_descriptors.total()*_descriptors.elemSize();
}
void SensorData::clearCompressedData(bool images, bool scan, bool userData)
{
if(images)
{
_imageCompressed=cv::Mat();
_depthOrRightCompressed=cv::Mat();
}
if(scan)
{
_laserScanCompressed.clear();
}
if(userData)
{
_userDataCompressed=cv::Mat();
}
}
void SensorData::clearRawData(bool images, bool scan, bool userData)
{
if(images)
{
_imageRaw=cv::Mat();
_depthOrRightRaw=cv::Mat();
}
if(scan)
{
_laserScanRaw.clear();
}
if(userData)
{
_userDataRaw=cv::Mat();
}
}
bool SensorData::isPointVisibleFromCameras(const cv::Point3f & pt) const
{
if(_cameraModels.size() >= 1)

8
corelib/src/odometry/OdometryF2M.cpp
View File

@@ -957,7 +957,7 @@ Transform OdometryF2M::computeTransform(
if(mapScan.is2d())
{
map_->sensorData().setLaserScanRaw(
map_->sensorData().setLaserScan(
LaserScan(
mapScan.data(),
0,
@@ -967,7 +967,7 @@ Transform OdometryF2M::computeTransform(
}
else
{
map_->sensorData().setLaserScanRaw(
map_->sensorData().setLaserScan(
LaserScan(
mapScan.data(),
0,
@@ -1137,7 +1137,7 @@ Transform OdometryF2M::computeTransform(
if(lastFrame_->sensorData().laserScanRaw().is2d())
{
Transform mapViewpoint(-newFramePose.x(), -newFramePose.y(),0,0,0,0);
map_->sensorData().setLaserScanRaw(
map_->sensorData().setLaserScan(
LaserScan(
util3d::laserScan2dFromPointCloud(*mapCloudNormals, mapViewpoint),
0,
@@ -1148,7 +1148,7 @@ Transform OdometryF2M::computeTransform(
else
{
Transform mapViewpoint(-newFramePose.x(), -newFramePose.y(), -newFramePose.z(),0,0,0);
map_->sensorData().setLaserScanRaw(
map_->sensorData().setLaserScan(
LaserScan(
util3d::laserScanFromPointCloud(*mapCloudNormals, mapViewpoint),
0,

58
corelib/src/optimizer/OptimizerG2O.cpp
View File

@@ -60,6 +60,8 @@ typedef Eigen::Matrix<double,Eigen::Dynamic,Eigen::Dynamic,Eigen::ColMajor> Matr
#include "g2o/types/slam2d/types_slam2d.h"
#include "g2o/types/slam3d/types_slam3d.h"
#include "g2o/edge_se3_xyzprior.h"
#include "g2o/edge_se3_gravity.h"
#include "g2o/edge_sbacam_gravity.h"
#ifdef G2O_HAVE_CSPARSE
#include "g2o/solvers/csparse/linear_solver_csparse.h"
#endif
@@ -306,7 +308,7 @@ std::map<int, Transform> OptimizerG2O::optimize(
{
for(std::multimap<int, Link>::const_iterator iter=edgeConstraints.begin(); iter!=edgeConstraints.end(); ++iter)
{
if(iter->second.from() == iter->second.to())
if(iter->second.from() == iter->second.to() && iter->second.type() == Link::kPosePrior)
{
rootId = 0;
break;
@@ -494,6 +496,7 @@ std::map<int, Transform> OptimizerG2O::optimize(
1 / static_cast<double>(iter->second.infMatrix().at<double>(4,4)) >= 9999.0 ||
1 / static_cast<double>(iter->second.infMatrix().at<double>(5,5)) >= 9999.0)
{
//GPS XYZ case
EdgeSE3XYZPrior * priorEdge = new EdgeSE3XYZPrior();
g2o::VertexSE3* v1 = (g2o::VertexSE3*)optimizer.vertex(id1);
priorEdge->setVertex(0, v1);
@@ -517,6 +520,7 @@ std::map<int, Transform> OptimizerG2O::optimize(
}
else
{
// XYZ+RPY case
g2o::EdgeSE3Prior * priorEdge = new g2o::EdgeSE3Prior();
g2o::VertexSE3* v1 = (g2o::VertexSE3*)optimizer.vertex(id1);
priorEdge->setVertex(0, v1);
@@ -536,6 +540,25 @@ std::map<int, Transform> OptimizerG2O::optimize(
}
}
}
else if(!isSlam2d() && gravitySigma() > 0 && iter->second.type() == Link::kPoseOdom && poses.find(iter->first) != poses.end())
{
Eigen::Matrix<double, 6, 1> m;
// Up vector in robot frame
m.head<3>() = Eigen::Vector3d::UnitZ();
// Observed Gravity vector in world frame
float roll, pitch, yaw;
iter->second.transform().getEulerAngles(roll, pitch, yaw);
m.tail<3>() = Transform(0,0,0,roll,pitch,0).toEigen3d() * -Eigen::Vector3d::UnitZ();
Eigen::MatrixXd information = Eigen::MatrixXd::Identity(3, 3) * 1.0/(gravitySigma()*gravitySigma());
g2o::VertexSE3* v1 = (g2o::VertexSE3*)optimizer.vertex(id1);
EdgeSE3Gravity* priorEdge(new EdgeSE3Gravity());
priorEdge->setMeasurement(m);
priorEdge->setInformation(information);
priorEdge->vertices()[0] = v1;
edge = priorEdge;
}
}
else if(id1<0 || id2 < 0)
{
@@ -1386,7 +1409,38 @@ std::map<int, Transform> OptimizerG2O::optimizeBA(
int id1 = iter->second.from();
int id2 = iter->second.to();
if(id1 != id2) // not supporting prior
if(id1 == id2)
{
#ifndef RTABMAP_ORB_SLAM2
g2o::HyperGraph::Edge * edge = 0;
if(gravitySigma() > 0 && iter->second.type() == Link::kPoseOdom && poses.find(iter->first) != poses.end())
{
Eigen::Matrix<double, 6, 1> m;
// Up vector in robot frame
m.head<3>() = Eigen::Vector3d::UnitZ();
// Observed Gravity vector in world frame
float roll, pitch, yaw;
iter->second.transform().getEulerAngles(roll, pitch, yaw);
m.tail<3>() = Transform(0,0,0,roll,pitch,0).toEigen3d() * -Eigen::Vector3d::UnitZ();
Eigen::MatrixXd information = Eigen::MatrixXd::Identity(3, 3) * 1.0/(gravitySigma()*gravitySigma());
g2o::VertexCam* v1 = (g2o::VertexCam*)optimizer.vertex(id1);
EdgeSBACamGravity* priorEdge(new EdgeSBACamGravity());
priorEdge->setMeasurement(m);
priorEdge->setInformation(information);
priorEdge->vertices()[0] = v1;
edge = priorEdge;
}
if (edge && !optimizer.addEdge(edge))
{
delete edge;
UERROR("Map: Failed adding constraint between %d and %d, skipping", id1, id2);
return optimizedPoses;
}
#endif
}
else if(id1>0 && id2>0) // not supporting landmarks
{
UASSERT(!iter->second.transform().isNull());

100
corelib/src/optimizer/OptimizerGTSAM.cpp
View File

@@ -50,7 +50,9 @@ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#include <gtsam/nonlinear/NonlinearOptimizer.h>
#include <gtsam/nonlinear/Marginals.h>
#include <gtsam/nonlinear/Values.h>
#include "optimizer/gtsam/GravityFactor.h"
#include "gtsam/GravityFactor.h"
#include "gtsam/GPSPose2XYFactor.h"
#include "gtsam/GPSPose3XYZFactor.h"
#ifdef RTABMAP_VERTIGO
#include "vertigo/gtsam/betweenFactorMaxMix.h"
@@ -104,14 +106,27 @@ std::map<int, Transform> OptimizerGTSAM::optimize(
gtsam::NonlinearFactorGraph graph;
// detect if there is a global pose prior set, if so remove rootId
bool gpsPriorOnly = false;
if(!priorsIgnored())
{
for(std::multimap<int, Link>::const_iterator iter=edgeConstraints.begin(); iter!=edgeConstraints.end(); ++iter)
{
if(iter->second.from() == iter->second.to())
if(iter->second.from() == iter->second.to() && iter->second.type() == Link::kPosePrior)
{
rootId = 0;
break;
if ((isSlam2d() && 1 / static_cast<double>(iter->second.infMatrix().at<double>(5,5)) < 9999) ||
(1 / static_cast<double>(iter->second.infMatrix().at<double>(3,3)) < 9999.0 &&
1 / static_cast<double>(iter->second.infMatrix().at<double>(4,4)) < 9999.0 &&
1 / static_cast<double>(iter->second.infMatrix().at<double>(5,5)) < 9999.0))
{
// orientation is set, don't set root prior
gpsPriorOnly = false;
rootId = 0;
break;
}
else if(gravitySigma()<=0)
{
gpsPriorOnly = true;
}
}
}
}
@@ -128,7 +143,11 @@ std::map<int, Transform> OptimizerGTSAM::optimize(
}
else
{
gtsam::noiseModel::Diagonal::shared_ptr priorNoise = gtsam::noiseModel::Diagonal::Variances((gtsam::Vector(6) << 1e-6, 1e-6, 1e-6, 1e-4, 1e-4, 1e-4).finished());
gtsam::noiseModel::Diagonal::shared_ptr priorNoise = gtsam::noiseModel::Diagonal::Variances(
(gtsam::Vector(6) <<
(gpsPriorOnly?2:1e-2), gpsPriorOnly?2:1e-2, gpsPriorOnly?2:1e-2,
1e-2, 1e-2, 1e-2
).finished());
graph.add(gtsam::PriorFactor<gtsam::Pose3>(rootId, gtsam::Pose3(initialPose.toEigen4d()), priorNoise));
}
}
@@ -207,42 +226,65 @@ std::map<int, Transform> OptimizerGTSAM::optimize(
{
if(isSlam2d())
{
Eigen::Matrix<double, 3, 3> information = Eigen::Matrix<double, 3, 3>::Identity();
if(!isCovarianceIgnored())
if (1 / static_cast<double>(iter->second.infMatrix().at<double>(5,5)) >= 9999.0)
{
information(0,0) = iter->second.infMatrix().at<double>(0,0); // x-x
information(0,1) = iter->second.infMatrix().at<double>(0,1); // x-y
information(0,2) = iter->second.infMatrix().at<double>(0,5); // x-theta
information(1,0) = iter->second.infMatrix().at<double>(1,0); // y-x
information(1,1) = iter->second.infMatrix().at<double>(1,1); // y-y
information(1,2) = iter->second.infMatrix().at<double>(1,5); // y-theta
information(2,0) = iter->second.infMatrix().at<double>(5,0); // theta-x
information(2,1) = iter->second.infMatrix().at<double>(5,1); // theta-y
information(2,2) = iter->second.infMatrix().at<double>(5,5); // theta-theta
noiseModel::Diagonal::shared_ptr model = noiseModel::Diagonal::Variances(Vector2(
1/iter->second.infMatrix().at<double>(0,0),
1/iter->second.infMatrix().at<double>(1,1)));
graph.add(GPSPose2XYFactor(id1, gtsam::Point2(iter->second.transform().x(), iter->second.transform().y()), model));
}
else
{
Eigen::Matrix<double, 3, 3> information = Eigen::Matrix<double, 3, 3>::Identity();
if(!isCovarianceIgnored())
{
information(0,0) = iter->second.infMatrix().at<double>(0,0); // x-x
information(0,1) = iter->second.infMatrix().at<double>(0,1); // x-y
information(0,2) = iter->second.infMatrix().at<double>(0,5); // x-theta
information(1,0) = iter->second.infMatrix().at<double>(1,0); // y-x
information(1,1) = iter->second.infMatrix().at<double>(1,1); // y-y
information(1,2) = iter->second.infMatrix().at<double>(1,5); // y-theta
information(2,0) = iter->second.infMatrix().at<double>(5,0); // theta-x
information(2,1) = iter->second.infMatrix().at<double>(5,1); // theta-y
information(2,2) = iter->second.infMatrix().at<double>(5,5); // theta-theta
}
gtsam::noiseModel::Gaussian::shared_ptr model = gtsam::noiseModel::Gaussian::Information(information);
graph.add(gtsam::PriorFactor<gtsam::Pose2>(id1, gtsam::Pose2(iter->second.transform().x(), iter->second.transform().y(), iter->second.transform().theta()), model));
gtsam::noiseModel::Gaussian::shared_ptr model = gtsam::noiseModel::Gaussian::Information(information);
graph.add(gtsam::PriorFactor<gtsam::Pose2>(id1, gtsam::Pose2(iter->second.transform().x(), iter->second.transform().y(), iter->second.transform().theta()), model));
}
}
else
{
Eigen::Matrix<double, 6, 6> information = Eigen::Matrix<double, 6, 6>::Identity();
if(!isCovarianceIgnored())
if (1 / static_cast<double>(iter->second.infMatrix().at<double>(3,3)) >= 9999.0 ||
1 / static_cast<double>(iter->second.infMatrix().at<double>(4,4)) >= 9999.0 ||
1 / static_cast<double>(iter->second.infMatrix().at<double>(5,5)) >= 9999.0)
{
memcpy(information.data(), iter->second.infMatrix().data, iter->second.infMatrix().total()*sizeof(double));
noiseModel::Diagonal::shared_ptr model = noiseModel::Diagonal::Precisions(Vector3(
iter->second.infMatrix().at<double>(0,0),
iter->second.infMatrix().at<double>(1,1),
iter->second.infMatrix().at<double>(2,2)));
graph.add(GPSPose3XYZFactor(id1, gtsam::Point3(iter->second.transform().x(), iter->second.transform().y(), iter->second.transform().z()), model));
}
else
{
Eigen::Matrix<double, 6, 6> information = Eigen::Matrix<double, 6, 6>::Identity();
if(!isCovarianceIgnored())
{
memcpy(information.data(), iter->second.infMatrix().data, iter->second.infMatrix().total()*sizeof(double));
}
Eigen::Matrix<double, 6, 6> mgtsam = Eigen::Matrix<double, 6, 6>::Identity();
mgtsam.block(0,0,3,3) = information.block(3,3,3,3); // cov rotation
mgtsam.block(3,3,3,3) = information.block(0,0,3,3); // cov translation
mgtsam.block(0,3,3,3) = information.block(0,3,3,3); // off diagonal
mgtsam.block(3,0,3,3) = information.block(3,0,3,3); // off diagonal
gtsam::SharedNoiseModel model = gtsam::noiseModel::Gaussian::Information(mgtsam);
Eigen::Matrix<double, 6, 6> mgtsam = Eigen::Matrix<double, 6, 6>::Identity();
mgtsam.block(0,0,3,3) = information.block(3,3,3,3); // cov rotation
mgtsam.block(3,3,3,3) = information.block(0,0,3,3); // cov translation
mgtsam.block(0,3,3,3) = information.block(0,3,3,3); // off diagonal
mgtsam.block(3,0,3,3) = information.block(3,0,3,3); // off diagonal
gtsam::SharedNoiseModel model = gtsam::noiseModel::Gaussian::Information(mgtsam);
graph.add(gtsam::PriorFactor<gtsam::Pose3>(id1, gtsam::Pose3(iter->second.transform().toEigen4d()), model));
graph.add(gtsam::PriorFactor<gtsam::Pose3>(id1, gtsam::Pose3(iter->second.transform().toEigen4d()), model));
}
}
}
else if(gravitySigma() > 0 && iter->second.type() == Link::kPoseOdom && poses.find(iter->first) != poses.end())
else if(!isSlam2d() && gravitySigma() > 0 && iter->second.type() == Link::kPoseOdom && poses.find(iter->first) != poses.end())
{
Vector3 r = gtsam::Pose3(iter->second.transform().toEigen4d()).rotation().xyz();
gtsam::Unit3 nG = gtsam::Rot3::RzRyRx(r.x(), r.y(), 0).rotate(gtsam::Unit3(0,0,-1));

88
corelib/src/optimizer/g2o/edge_sbacam_gravity.h Normal file
View File

@@ -0,0 +1,88 @@
/*
Copyright (c) 2010-2019, 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.
*/
/**
* Adapted from EdgeSE3Gravity
*/
#ifndef RTAB_G2O_EDGE_SBACAM_GRAVITY_H_
#define RTAB_G2O_EDGE_SBACAM_GRAVITY_H_
#include "g2o/types/sba/types_sba.h"
#include "g2o/core/base_unary_edge.h"
namespace rtabmap {
/**
* \brief EdgeSBACamGravity
* \brief g2o edge with gravity constraint
*/
class EdgeSBACamGravity : public g2o::BaseUnaryEdge<3, Eigen::Matrix<double, 6, 1>, g2o::VertexCam> {
public:
EIGEN_MAKE_ALIGNED_OPERATOR_NEW
EdgeSBACamGravity(){
information().setIdentity();
}
virtual bool read(std::istream& is) {return false;} // not implemented
virtual bool write(std::ostream& os) const {return false;} // not implemented
// return the error estimate as a 3-vector
void computeError(){
const g2o::VertexCam* v1 = static_cast<const g2o::VertexCam*>(_vertices[0]);
Eigen::Vector3d direction = _measurement.head<3>();
Eigen::Vector3d measurement = _measurement.tail<3>();
Eigen::Vector3d ea;
Eigen::Matrix3d t = v1->estimate().rotation().toRotationMatrix();
ea[0] = atan2(t (2, 1), t (2, 2));
ea[1] = asin(-t (2, 0));
ea[2] = atan2(t (1, 0), t (0, 0));
Eigen::Matrix3d rot =
(Eigen::AngleAxisd(ea[1], Eigen::Vector3d::UnitY()) *
Eigen::AngleAxisd(ea[0], Eigen::Vector3d::UnitX())).toRotationMatrix();
Eigen::Vector3d estimate = rot * -direction;
_error = estimate - measurement;
//printf("%d : measured=%f %f %f est=%f %f %f error=%f %f %f\n", v1->id(),
// measurement[0], measurement[1], measurement[2],
// estimate[0], estimate[1], estimate[2],
// _error[0], _error[1], _error[2]);
}
// 6 values:
// [0:2] Up vector in robot frame
// [3:5] Observed gravity vector in world frame
virtual void setMeasurement(const Eigen::Matrix<double, 6, 1>& m){
_measurement.head<3>() = m.head<3>().normalized();
_measurement.tail<3>() = m.tail<3>().normalized();
}
};
}
#endif

89
corelib/src/optimizer/g2o/edge_se3_gravity.h Normal file
View File

@@ -0,0 +1,89 @@
/*
Copyright (c) 2010-2019, 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.
*/
/**
* Adapted code from HDL graph slam:
* https://github.com/koide3/hdl_graph_slam/blob/master/include/g2o/edge_se3_priorvec.hpp
*/
#ifndef RTAB_G2O_EDGE_SE3_GRAVITY_H_
#define RTAB_G2O_EDGE_SE3_GRAVITY_H_
#include "g2o/core/base_unary_edge.h"
#include "g2o/types/slam3d/vertex_se3.h"
namespace rtabmap {
/*! \class EdgeSE3Gravity
* \brief g2o edge with gravity constraint
*/
class EdgeSE3Gravity : public g2o::BaseUnaryEdge<3, Eigen::Matrix<double, 6, 1>, g2o::VertexSE3> {
public:
EIGEN_MAKE_ALIGNED_OPERATOR_NEW
EdgeSE3Gravity(){
information().setIdentity();
}
virtual bool read(std::istream& is) {return false;} // not implemented
virtual bool write(std::ostream& os) const {return false;} // not implemented
// return the error estimate as a 3-vector
void computeError(){
const g2o::VertexSE3* v1 = static_cast<const g2o::VertexSE3*>(_vertices[0]);
Eigen::Vector3d direction = _measurement.head<3>();
Eigen::Vector3d measurement = _measurement.tail<3>();
Eigen::Vector3d ea;
Eigen::Matrix3d t = v1->estimate().linear();
ea[0] = atan2(t (2, 1), t (2, 2));
ea[1] = asin(-t (2, 0));
ea[2] = atan2(t (1, 0), t (0, 0));
Eigen::Matrix3d rot =
(Eigen::AngleAxisd(ea[1], Eigen::Vector3d::UnitY()) *
Eigen::AngleAxisd(ea[0], Eigen::Vector3d::UnitX())).toRotationMatrix();
Eigen::Vector3d estimate = rot * -direction;
_error = estimate - measurement;
//printf("%d : measured=%f %f %f est=%f %f %f error=%f %f %f\n", v1->id(),
// measurement[0], measurement[1], measurement[2],
// estimate[0], estimate[1], estimate[2],
// _error[0], _error[1], _error[2]);
}
// 6 values:
// [0:2] Up vector in robot frame
// [3:5] Observed gravity vector in world frame
virtual void setMeasurement(const Eigen::Matrix<double, 6, 1>& m){
_measurement.head<3>() = m.head<3>().normalized();
_measurement.tail<3>() = m.tail<3>().normalized();
}
};
}
#endif

4
corelib/src/optimizer/g2o/edge_se3_xyzprior.cpp
View File

@@ -87,10 +87,6 @@ void EdgeSE3XYZPrior::computeError() {
_error = v->estimate().translation() - _measurement;
}
void EdgeSE3XYZPrior::linearizeOplus() {
_jacobianOplusXi << Eigen::Matrix3d::Identity();
}
bool EdgeSE3XYZPrior::setMeasurementFromState() {
const g2o::VertexSE3* v = static_cast<const g2o::VertexSE3*>(_vertices[0]);
_measurement = v->estimate().translation();

1
corelib/src/optimizer/g2o/edge_se3_xyzprior.h
View File

@@ -63,7 +63,6 @@ public:
virtual bool read(std::istream& is);
virtual bool write(std::ostream& os) const;
virtual void computeError();
virtual void linearizeOplus();
virtual bool setMeasurementFromState();
virtual double initialEstimatePossible(const g2o::OptimizableGraph::VertexSet& /*from*/, g2o::OptimizableGraph::Vertex* /*to*/) {return 1.;}

57
corelib/src/optimizer/gtsam/GPSPose2XYFactor.h Normal file
View File

@@ -0,0 +1,57 @@
/**
* Author: Mathieu Labbe
* This file is a copy of GPSPose2Factor.h of gtsam examples
*/
/**
* A simple 2D 'GPS' like factor
* The factor contains a X-Y position measurement (mx, my) for a Pose, but no rotation information
* The error vector will be [x-mx, y-my]'
*/
#pragma once
#include <gtsam/nonlinear/NonlinearFactor.h>
#include <gtsam/base/Matrix.h>
#include <gtsam/base/Vector.h>
#include <gtsam/geometry/Pose2.h>
namespace rtabmap {
class GPSPose2XYFactor: public gtsam::NoiseModelFactor1<gtsam::Pose2> {
private:
// measurement information
double mx_, my_;
public:
/**
* Constructor
* @param poseKey associated pose varible key
* @param model noise model for GPS snesor, in X-Y
* @param m Point2 measurement
*/
GPSPose2XYFactor(gtsam::Key poseKey, const gtsam::Point2 m, gtsam::SharedNoiseModel model) :
gtsam::NoiseModelFactor1<gtsam::Pose2>(model, poseKey), mx_(m.x()), my_(m.y()) {}
// error function
// @param p the pose in Pose2
// @param H the optional Jacobian matrix, which use boost optional and has default null pointer
gtsam::Vector evaluateError(const gtsam::Pose2& p, boost::optional<gtsam::Matrix&> H = boost::none) const {
// note that use boost optional like a pointer
// only calculate jacobian matrix when non-null pointer exists
if (H) *H = (gtsam::Matrix23() << 1.0, 0.0, 0.0,
0.0, 1.0, 0.0).finished();
// return error vector
return (gtsam::Vector2() << p.x() - mx_, p.y() - my_).finished();
}
};
} // namespace gtsamexamples

53
corelib/src/optimizer/gtsam/GPSPose3XYZFactor.h Normal file
View File

@@ -0,0 +1,53 @@
/**
* Author: Mathieu Labbe
* This file is a copy of GPSPose2Factor.h of gtsam examples for Pose3
*/
/**
* A simple 3D 'GPS' like factor
* The factor contains a X-Y-Z position measurement (mx, my, mz) for a Pose, but no rotation information
* The error vector will be [x-mx, y-my, z-mz]'
*/
#pragma once
#include <gtsam/nonlinear/NonlinearFactor.h>
#include <gtsam/base/Matrix.h>
#include <gtsam/base/Vector.h>
#include <gtsam/geometry/Pose3.h>
namespace rtabmap {
class GPSPose3XYZFactor: public gtsam::NoiseModelFactor1<gtsam::Pose3> {
private:
// measurement information
double mx_, my_, mz_;
public:
/**
* Constructor
* @param poseKey associated pose variable key
* @param model noise model for GPS sensor, in X-Y
* @param m Point2 measurement
*/
GPSPose3XYZFactor(gtsam::Key poseKey, const gtsam::Point3 m, gtsam::SharedNoiseModel model) :
gtsam::NoiseModelFactor1<gtsam::Pose3>(model, poseKey), mx_(m.x()), my_(m.y()), mz_(m.z()) {}
// error function
// @param p the pose in Pose
// @param H the optional Jacobian matrix, which use boost optional and has default null pointer
gtsam::Vector evaluateError(const gtsam::Pose3& p, boost::optional<gtsam::Matrix&> H = boost::none) const {
if(H)
{
p.translation(H);
}
return (gtsam::Vector3() << p.x() - mx_, p.y() - my_, p.z() - mz_).finished();
}
};
} // namespace gtsamexamples

7
corelib/src/optimizer/gtsam/GravityFactor.cpp
View File

@@ -9,6 +9,13 @@
* -------------------------------------------------------------------------- */
/**
* Author: Mathieu Labbe
* This file is a copy of AttitudeFactor.cpp of gtsam library but
* with attitudeError() function overridden to ignore yaw errors.
* For the noise model, use Sigmas(Vector2(0.1, 10)) (with second sigma high!)
*/
/**
* @file GravityFactor.cpp
* @author Frank Dellaert