4 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
3 people	106874845f	CuVSLAM 14 Update (#1618) ... * update cuvslam api call to adhere to new v4.0 changes * update comment * parse cuvslam major version from the header and throw an error if theversion doesn't match the expected * moving average filter + extensive debug logging * lots more logs * improvement by turning off use_motion_model in cuvslam config. lots of debug logs. * slight improvements to velocity guard logic. lots of debug logs * removing debug logs. working. * let find_package_handle_standard_args handle the version comparison logic. Custom logic to handle newer then requested major version * proper warning message if the major version doesn't match * add back debug logs. Handle no guess transform case. * adding warnings when no guess transform is provided * clean up comments * lowering average window to 5 * moderate odometry mode, 5 value averaging window * adding comment * add reset logs to investigate reset issue * try returning guess or identity on init to avoid reset cascade * disable motion model again * big log when tracking * remove some debug logs * flag to use original covariance from cuVSLAM * use raw covariance * data collection for revised lost detection * compare guess to estimated transform for lost detection * re-enable motion model, it appears to reduce false-positives for the new lost detection * run pose estimation afer init on first frame * don't initialize if we don't have enough features * dont warm up GPU everytime we reset * low estimate velocity guard. Passes all challenges. Debug comments still everywhere. * Remove debug logs. Add thresholds to header in a config section. * Cleanup logs * exposing multi-cam mode as a rtabmap param * remove max frame delta * change to use assertion instead of conditional * replaced some UERROR by UWARN --------- Co-authored-by: Felix Toft <[email protected]> Co-authored-by: matlabbe <[email protected]>	2025-12-15 14:21:32 -08:00
Felix ToftandFelix Toft	d1b5d62d21	Limit Max Features (#1614) ... * limit max features internally inside pydetector and superpoint_rpautrat to limit keypoint/desc data size and improve performance * avoid full reset when max features is changed to work around the memory temprorary param update logic * remove leftover comment * try using image roi instead * revert and regenerate * unintended change * fixing issue with pydetector, adding mask filtering --------- Co-authored-by: Felix Toft <[email protected]>	2025-11-16 11:37:03 -08:00
Felix ToftandFelix Toft	5e4fd171e2	SuperPoint Rpautrat (MIT license) (#1603) ... * initial python implementation of superpoint rpautrat * working python implementation of superpoint * small tweaks to try and work around the GIL issue * fix missing os import * begging cpp impl of superpoint python model * finishing cpp superpoint impl using the same SPDetector interface * finalizing superpoint cpp impl, working with cpu but needs to be cleaned * fixing feature matching by reworking nms and filtering logic * speeding up nms with batched operations and cleaning up * updating conversion script * adding args for image dimensions and cuda usage to model tracer * wiring up UI to parse superpoint params * adding label to superpoint rpautrat ui * reverting some unintended ui changes * typo * oneline revert * removing nms and threshold filtering from cpp, this is handled internally by the superpoint model * using python interface to run the superpoint _to_torchscript.py script at runtime * generate and load model file on the first incoming frame * reverting unintented change * ui changes appear mysteriously again, reverting * remove topk from model to prevent issue when number of detected keypoints is lower then the k value (scripting fails in this case) * cleaning up for review * change dest for model file and remove debug logs * bump patch and add version comment * rm unucessary comment * use resources to load file to ensure it works when rtabmap isn't built from source * execute with pybind runpy instead of system call * only build superpoint rpautrat if we have torch and python support * changing param to accept a path to the weights .pth file directly * parse the default working directory to save the model file in * rm extraneous change * rm setters and re-initialize the detector whenever params change. We cannot support changing params after the model is constructed * update UI text * only enable superpoint rpautrat when built with python and torch * more build information regarding superpoint rpautrat * generate temporary python script in the working dir * rm unecessary changes to rtabmap_superpoint.py * fix comment and only add repo root to sys.path * introduce a new parameter for the superpoint python model definition * execute script from string instead of writing to a file * wrap parse params in a single compiler directive * remove descriptor spatial matching logic and rely on upstream RTAB-Map processes to take the top-K desc and kpts * only add python script to resources if built with superpoint rpautrat support. supress warning with type casting * isolate pybind11 setup so it can't affect any other modules or potential regnerations of the model file * update about dialog to show superpoint rpautrat * remove debug logs --------- Co-authored-by: Felix Toft <[email protected]>	2025-11-07 17:22:35 -08:00
3 people	5aec9bacec	CuVSLAM VO Strategy (#1583) ... * integrating cuvslam, stereo cam initialization * fixing transformations * use isaac ros TocuVSLAMPose * organizing * handling covariance conversion * cleaning up config * moving cmake instructions to FindCuVSLAM.cmake file * fixing ui integration * updating software license for cuvslam * improving memory management * cleaning up header more * reverting mistake in GUI * using opaque pointers for header definitions, compiles * back to typed definitions using forward-referencing * fixing dangling pointer * cleaning up * fixing variable scope issues * Using cuda to allocate gpu memory * fixing transfrom from cuvslam back to ros coordinate frame * reducing excessive logs * removing redundant if true in cuvslam cmake * moving find cuda into cuvslam cmake * cleaning cmake to use modern target * adding copyright line * Finding tf2 and eigen3 in cmake * simplifying clean up logic * better logs * moving cuvslam implementation methods into cpp * fixing build with implmentation function declarations moved to the cpp * cleaning transformation logic and adding grayscale support * typo * removing unecessary class member variable for processed images * remove trailing underscore from initialize cuvslam function signature * locally scoping config params since they are copied by cuvslam * removing unecessary member variables and using locally scoped values since cuvslam seems to copy them * more cleaning and refactoring * fixing cmake inconsistancies and surpress warning * fixing up baseline transform * trying to remove tf2 for transformations * fixing incorect baseline transform swap * comment, begin reset logic investigation * better handling of previous pose and transform during error cases * hunting for reset bug, not found * observation export implementatin, currently getting strange errors * multi cam support * Memory fixes * comments * make cuda stream a member variable so it isn't initialized and destroyed on every frame * preparing to add ground constraints * adding ground constraints, working * adding wheel odom fusion into cuvslam tracker * parsing parameter to apply planar constraints * removing planar constraint in config, doesn't do anything * using proper cuda stream type for member variable * cleaning up for merge * only parse params when cuvslam is compiled * updating error message if using rgb-d instead of stereo * removing wheel odom testing logs * reverting bgr to rgb conversion * fixing initial pose logic * return null transform on invalid covariance * debugging covariance * guard against low velocity situations where cuvslam covariance spikes, but we aren't lost * cleaning things up --------- Co-authored-by: Felix Toft <[email protected]> Co-authored-by: matlabbe <[email protected]>	2025-10-03 11:49:21 -07:00