arjo129 / darpa_ugv_highlevel Goto Github PK

• Get decent X,Y wheel odometry for OGV (apply differential param on robot_localization)
• Publish covariance values for rf2o odometry
• Merge realsense orientation, wheel odom (with covariance) and rf2o laser odom

Realsense drivers and ros package has issues with latest kernel version (v5.0 and above). Downgrade all 5 Udoo ubuntu kernels with immediate effect to improve pointcloud processing (30fps on old kernel, 5fps on new kernel) and stability.

Does not require re-flashing or re-installing as just need to allow Ubuntu to boot from the latest kernel, which should be v4.15 after removal.

Also do check if this kernel version messes up with other packages to prevent last minute panic.

• Add alert for WiFi sensor, possibly add marker on map
• Add alert for CO2 Reading, possibly add marker on map
• LaserScan from 1 robot with TF (DARPA Frame)
• View robot current pose on map (DARPA to base_link frame)
• Multi-Robot maps (separate mapping panes for each robot)
• Multi-Robot maps final merge (merge all the maps together for 1 coherent map)
• Location of WiFi markers (poop trail) dropped by UGV

• Fix path planner (dwa planner) generating paths around obstacles with narrow gaps without considering footprint and path hugging around obstacles

• Fix repeated forward and reverse motion to drive around vehicles despite high rotation capability of OGV

• Fix frequent switching of frontier goals given due to long time required to reach goal in presence of cluttered obstacle field

• Add drop-down and XYZ coordinate (DARPA frame) box to allow reporting of Artifacts ("Survivor", "Backpack", "Cell Phone", "Vent", "Gas")
• Use Interface_Protocol package rosservice to report XYZ coordinate (DARPA frame) and artifact name.

Teensy is not seen in the udoo (under any of the /dev ports) and teensy_bridge package cannot work. Likely due to USB hub usage. Also need to test thermal camera, xbee and LoRa integration on the teensy.

Quite urgent as difficult to do transform between realsense and thermal camera for sensor fusion without URDF. Floor frame also required (different frame for OGV and TGV) for initial DARPA frame transform.

• OGV URDF with floor frame
• TGV URDF with floor frame (by hand)

data_compressor package

• Multi-robot goal pose (visual markers to aid goal)
• Multi-robot estop
• Multi-robot start (undo e-stop)

Upon subscribing to the pointcloud topic of the realsense, it continuously throws errors about "out of frame resources" and other errors. Frame rate drops from 30fps to 5fps and affects mapping. Sometimes replicable on a desktop/laptop, but not as error prone as the Udoo.

• Add pane to allow viewing of score, current time, number of REPORTS left (artifacts) and verify current team name
• Interface_Protocol package to publish this status report once every 10s (configurable to prevent flooding and banning from DARPA server)

RTAB map is able to trigger when it has lost track of the environment to allow us to restart the RTAB map node. However currently unsure of how to restart at a decent estimate of a current point based on old RTAB map and the predicted odom readings

• Get RTAB map to publish map -> odom TF correction frame
• Restart RTAB map at sensible estimated pose