It is checked if the trajectory is kinematically feasible and collides with the obstacles.
IMHO, this should be reviewed if locally it is detected that the trajectory is feasible and does not collide with anything. Personally I have no idea about the web script. Perhaps, send the solution by email to the commonroad email with topic [Scenario problem] and ask to review how this particular scenario is handled by the web script if locally it causes no problem. This scenario should work. After the deadline, all solutions will be properly checked but it’s really important feedback that there are any problems.
Please check if the solution works locally and if submission of other solutions works.
I got the similar problem: I found a solution from the jupyter notebook, bur when I upload to the website, it says not feasible. When I visualize the result, I found the ego vehicle may be too close to the front car, but no collision. Will it be accepted in Commonroad?
It should be decided by the teaching team. It is very easy to programmatically detect and accept such solutions. The floating-point precision of double numbers is probably slightly lost when the trajectories are converted into the human-readable form of the xml file.
Are we allowed to use different vehical types in this programming exercise? i.e changing the veh_type from ‘‘BMW 320i’’ in the current jupyter notebook to eg. ‘‘FORD_ESCORT’’. By doing this I solved the problem that ‘‘Vehicle trajectory is not feasible for vehicle model KS2 in planning problem solution xxx’’. And the solution file will be written as ‘‘solution_KS1:SM1:xxx’’
I recommend that you increase the width and length of your car by a centimeter before search so that it avoids collision with other cars at least by a centimeter. I’m not sure allowing car crashes would be the best way to go for the public web script.
It takes exactly five minutes to figure out in which three files these two parameters can be changed.
Hi, also, one question which might be related to this, so I submitted some solutions twice, and is there w way to count how many I have done (uniquely) successfully published?
yes it does, only thing is I have published some of them 2x or 3x as I did not knew it inintially so they all show up irrespective of being the same solution in this!!
Yes thanks for ur idea!
and some errors can be solved in this way but some can still not.
Recommend donot to change the width to bigger because it will cause another “can not reach goal” error when submitting.
BTW the related file is MotionAutomate.py.
It now takes only five seconds to figure out in which file these two parameters can be changed.
I actually thought it was parameters_vehicle*.py because that’s what is used for Trajectory generation. But I have no idea.
In daily research, people apply minkowski sums, extending the obstacles by a radius.