search

utexas-bwi / eband_local_planner

ROS Local planner based on the eband approach. Based on the original implementation by Christian Connette and Bhaskara Marathi. This local planner has been adapted primarily for differential drive robots, but still supports the original holonomic drive controls.
http://wiki.ros.org/eband_local_planner
105 stars 83 forks source link

robot footprint #14

Closed aswinthomas closed 9 years ago

aswinthomas commented 10 years ago

Hi, Is the robot footprint considered when converting the plan to band? Or is this planner solely for a point robot?

I have a robot that is 1.7mx0.6m and it often hits walls.

piyushk commented 10 years ago

@aswinthomas The circumscribed radius from the costmap is considered when converting the plan to a band, so in principle it should work.

aswinthomas commented 10 years ago

Hi Piyush, Do I have to specify the "radius" parameter for costmap? the wall hit behavior happens when turning on spot.

jack-oquin commented 10 years ago

Yes.

The robot navigates as a point, but the costmap radius inflates obstacles so it should not hit them. For that to work, you need to measure the radius from the robot's axis of rotation.

aswinthomas commented 10 years ago

Hi Jack, Thank you. That works. Unfortunately this is not a good planner if you have a rectangular robot that is long with respective to its environment and the obstacles that it is expected to negotiate. You end up with many 'no solution' situations.

Otherwise, its actually a great planner. For true holonomic motions, you dont need to limit the translational control_deviation with respective to angular control_deviation as done in getTwist() in trajectory controller. I have yet to test it on an actual robot. Will update you guys.

jack-oquin commented 9 years ago

Can we close this issue?

Other than the comment that it does not work well with highly rectangular robots, it is not clear what problem is being reported.