v0.6.1 (genmos/ros2)

This release contains a ROS2 package for GenMOS (in ros2 folder) with a simulation test that works the same way as the simulation test of GenMOS in ROS1. After building this package within your ROS2 workspace (follow instructions here), you can run this simulation test by

ros2 run genmos_object_search_ros2 test_genmos_ros2_sim.sh

and a terminal window with multiple tabs should pop up, and an RViZ should also pop up that should show something like below (may need to wait a few seconds):

The ROS1 package for GenMOS (as in v0.5.1) should still work as that part of the codebase is not affected by this release.

No robot integration is attempted or planned with this ROS2 package, but it should be possible, similar to how the ROS1 integration works (see GenMOS on Spot for inspiration). Developers are welcome to attempt their own integration.

v0.5.1 (genmos)

The package is renamed to genmos_object_search; use GenMOS in accordance with the paper.

To run a test, check out ros/tests/spot/grpc/simple_sim/test_simple_sim_env_local_search_3d.py. Follow the instructions in the comments,

# To run the test, do the following IN ORDER:
# 0. run config_simple_sim_lab121_lidar.py to generate the .yaml configuration file
# 1. run in a terminal 'roslaunch rbd_spot_perception graphnav_map_publisher.launch map_name:=lab121_lidar'
# 2. run in a terminal 'roslaunch genmos_object_search_ros spot_local_cloud_publisher.launch robot_pose_topic:=/simple_sim_env/init_robot_pose'
# 3. run in a terminal 'roslaunch genmos_object_search_ros simple_sim_env.launch map_name:=lab121_lidar'
# 4. run in a terminal 'python -m genmos_object_search.grpc.server'
# 5. run in a terminal 'python test_simple_sim_env_local_search_3d.py groundtruth 32 config_simple_sim_lab121_lidar.yaml
# 6. run in a terminal 'roslaunch genmos_object_search_ros view_simple_sim.launch'
# 7. to monitor CPU temperature: 'watch -n 1 -x sensors'

This opens up RViZ (step 6) which should show something like:

This requires setting up robotdev for Spot (which contains rbd_spot_perception). This is because the SLAM map used in this test is built by Spot in our lab. Note that you don't need to have a Spot in order to setup the robotdev for Spot. Just follow: https://github.com/zkytony/robotdev/tree/master/spot

The ROS package inside this release is at the state when evaluation was conducted on Spot, MOVO and UR5e. This includes code, launch files & configurations used in those evaluations. Future releases may overwrite those for better organization.

v0.4.1-postviam

After viam work for UR5e. Before massive renaming (sloop_object_search --> genmos_object_search)

v0.3.5-beforeviam

repo state before viam work for UR5e robot

v0.1 (first base)

In this phase of work, we:

• Created the sloop_ros package with structure in place
• Refactor & Improve SLOOP system (in OSM 2D simulation)
  • In particular, sloop defines SLOOP as it should be (spatial language observation & spatial language observation model)
  • Ported over COS-POMDP's hierarchical planning algorithm over 2D
• Develop a robot-independent framework that integrates object search POMDP with ROS
• Verified sloop_ros with the Spot robot.