Debug

[bfalacerda]

Also kept the old params so we can do more structured comparison.

• acceleration limits changed to the same as used in mira
• min_trans_vel reduced, otherwise with the new acceleration limit the robot wont be able to sample any valid velocities
• decreased rot_stopped_vel and trans_stopped_vel so that the controller is more accurate when checking if the robot has stopped
• increased vx_samples from 3 to 10. since its a differential drive we need more on x to properly control the translational velocity
• decreased vy_samples to 1, as this robot doesnt move in y. In fact this doesn't make a difference in execution, because of the min_vel_y and max_vel_y being 0, it's just for coherence
• put sim_time back to default. with the increased vx_samples the control loop sometimes takes longer that it should so we reduce a bit of the computational burden here.
• changed forward_point_distance from 0.325 to 0.0, according to Making movebase work #62 (comment)

[nilsbore]

I did a few experiments with passing our doors today with forward_point_distance: 0.0. In the past we have had this problem that it's turning toward one side of the door when passing and this persists with this change. I'm gonna see if there might be any other benefits, if there are any negative impacts they should be been seen quite easily.

[bfalacerda]

i think we won't get rid of that without changing code

[marc-hanheide]

retest this please

[marc-hanheide]

there was a jenkins problem. It's tested now. I suggest to merge it, as this is helping to work towards #62 I guess?

[bfalacerda]

i think all the changes have a justification, that's why I did this PR to make them our starting point for testing. However, this is (again) just a bit of fiddling with params, with some unstructured testing. I kept the old values so we can run unit tests on both later on