local-planning-benchmark
local-planning-benchmark copied to clipboard
Example for collecting statistics
Thank you for your work. It compiled successfully on ros-noetic as well.
roslaunch move_base_benchmark move_base_benchmark.launch
and
roslaunch move_base_benchmark move_base_benchmark.launch
work as expected.
Could you provide some instructions on how to -change the local planner -collect statistics, in order to generate a table as table II in the paper?
Hi fabiodl,
For the first question, you can change the local planner by modifying move_base_benchmark.launch. For example, you can use dwa local planner instead of teb local planner via modifying
...
<param name="base_local_planner" value="teb_local_planner/TebLocalPlannerROS" />
<rosparam file="$(find move_base_benchmark)/params/teb_local_planner_params.yaml" command="load" />
...
to
...
<param name="base_local_planner" value="dwa_local_planner/DWAPlannerROS" />
<rosparam file="$(find move_base_benchmark)/params/dwa_local_planner_params.yaml" command="load" />
...
If you want to use a custom global or local planner, please refer to
Writing A Global Path Planner As Plugin in ROS
Writing A Local Path Planner As Plugin in ROS
For the second question, the intermediate data during the navigation process will be recorded in the log.txt file in move_base_benchmark/launch directory with the format of {timestamp, position_x, position_y, heading, linear_velocity, angular_velocity, distance_to_closest_obstacle, planning_time}. You can use the metrics.cpp script in move_base_benchmark/doc to compute some evaluation results as presented in our paper.
Thank you for your reply.
As for the tests, I believe the initial and goal poses are specified in the map10cm/*/config.txt files. However, these files report three start-goal pairs, while the paper's table lists five. Would it be possible to provide the remaining two positions?