-
Notifications
You must be signed in to change notification settings - Fork 1.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
if set goal from A to B, navigate will auto stop and robot not arrive expect point #926
Comments
@rafaelhuang-intel |
I have tried Ros2 core 0614 + Navigation2 / Dashing-devel (016df29), this issue still can reproduced. attached log and screen capture: or you can link here (https://drive.google.com/file/d/1_qkBAgfL-kzEUWiwhM2MakOmlB2z8SRQ/view?usp=sharing) to get video. |
@rafaelhuang-intel |
#936 is a possible workaround for this, I think. Looks like the robot is just getting too close to obstacles in the video |
An aside but something to bring up: collectively, a lot of time has been spent on DWB and it doesnt seem that anyone is happy with it. I propose we adapt it in the short term but talk about what it might look like to remove it entirely from support. MPC is a very popular technique for a reason and should be somewhat simple to generalize across different robot base types. The DRL work is interesting to @mhpanah and me but I recognize that even if I make it work to a degree I am happy with, its research code that wouldn't find its way to prime time for quite awhile until the RL community starts leveling out. Thoughts? |
I fully agree. I wanted to get the DWA/B algorithm approach to match the quality achieved in ROS 1 before spending time to find a better algorithm. There's no reason it shouldn't work as well or better than DWA/base_local_planner in ROS 1. If we can get a local planner interface defined like you mentioned in some other issue, that would be a good point to get all the ROS2 changes to DWB pushed back to @DLu's repo. Then navigation2 could just provide the core that DWB, TEB, MPC and others use. |
@crdelsey Highly in favor of such a local planner interface! |
Ticket here #943 Easier to make these changes now before trying to port multiple local or global planners. My recommendation would be to stop porting TEB until we have defined something unless someone actually needs it |
@yathartha3 has been looking into the port of TEB. Perhaps he can also look into this. |
If the changes are significant, I suggest we take some time to discuss the design and clarify the interface, integration components (plugins, adaptors, ROS wrapper), rethink role of a local planner (pure path following & collision avoidance?), and interaction with global planning (MPC can do both tasks) and set the stage for further work on enabling multi-robot. |
I have been testing Navigation2 (Dashing) with a real robot (Turtlebot 3). I almost never see the robot hitting obstacles or getting too close to them. @rafaelhuang-intel Do you still have this problem? |
I tested Navigation2 (Dashing) with a real robot (Turtlebot 3) & simulate by e-manual (http://emanual.robotis.com/docs/en/platform/turtlebot3/ros2/#ros2), I almost never see this issue, so I will close it. BTW, I see navigation 2 launch step not match dashing release yet(https://github.com/ros-planning/navigation2/tree/master/nav2_bringup), so do you have any plan to update this one? |
Bug report
I meet problem that if set goal from A to B, navigate will auto stop and robot not arrive expect point, but I can set goal again and robot workable except hit the post or wall, this issue meet rate ~30% ~ 40%, so do you meet same problem?
Required Info:
Steps to reproduce issue
Expected behavior
if set goal from A to B, robot arrive expect point
Actual behavior
if set goal from A to B, navigate will auto stop and robot not arrive expect point
Additional information
Feature request
Feature description
Implementation considerations
The text was updated successfully, but these errors were encountered: