planned and predicted turn radii differ

sjwo commented 3 years ago

[x] check speed
[x] check heading
[x] check turning radius. 8 m is recommended.
[x] Seeing better line-keeping behavior on initial path to start of ribbon on adjacent_multi_line when first manually setting PathPlanner instead of PathFollower in RQT. Thought that was automatic. Maybe not. Ok, my default_bit_star.scenario_config does explicitly say use planner 2 instead of default.scenario_config's planner 0. So I think I'm good on this, and must just have been making up this observation (there're obviously other things wrong that I am seeing anyway).
[x] disable all disturbances. Do this through a .scenario_config file?

sjwo commented 3 years ago

Maybe this is related to handling speed?

sjwo commented 3 years ago

Maybe heading is messed up?

sjwo commented 3 years ago

In Alex Walker's pydubins code, rho is described as "the turning radius of the vehicle."

sjwo commented 3 years ago

Ok, this is where I set speed in my BitStarPlanner.cpp:

DubinsWrapper dubins_wrapper = DubinsWrapper();
          // TODO figure out correct speed and start time to set in fill() call:
          dubins_wrapper.fill(
            dubins_path,
            1,
            start_time
          );

I think if I set this speed correctly, the dubins_wrapper.getEndTime() call will return the appropriate time, and that should make the controller happier (or at least less sad).

sjwo commented 3 years ago

sjwo / path_planner

planned and predicted turn radii differ #12