Changelog for package graceful_controller_ros

0.4.8 (2023-03-06)

fix stutter when robot velocity exceeds max_vel_x (#69) a bit of noise on the odometry can cause the robot to overshoot the max_vel_x - this would then cause us to decelerate - possibly quite hard if the decel_lim_x was high.
Contributors: Michael Ferguson

follow acceleration limits when velocity limit changes (#62)
add angular velocity limiting control (#61) Adds max_x_to_max_theta_scale_factor which can be used to control how the angular velocity changes relative to the linear velocity when use_vel_topic feature is enabled.
Contributors: Michael Ferguson, TrazCobalt

always rotate to match desired heading (#54) If the goal tolerances are large (greater than 0.5m) the robot might end up trying to point at the goal rather than aligning with the goal heading
Contributors: Michael Ferguson

use separate decel_lim_x for control law (#45)
add velocity limit to goal tolerance (#44)
fix deceleration time (#43) this reverts #6, to what is apparently the correct math
add footprint inflation for higher speeds (#32) * add a feature to inflate the footprint at higher speeds * only update max_vel_x once per call
initialze collision_points_ to null (#35)
add optional visualization of colliding points (#34)
split out hpp file, clang format (#33)
fix bug in latch_xy_goal_tolerance (#31) While the main control loop can latch the goal (thus it will stop trying to hunt around) - that isn't understood by the isGoalReached() function - which leads to the robot wiggling back and forth for a while
Contributors: Michael Ferguson

add ability to disable orientation filtering (#26) also make everything dynamic reconfigurable
don't crash when path is empty (#25)
add min_lookahead parameter (#24) when the pose is very close to the robot, we can get some instability (rapid side-to-side movement due to large angular errors over small linear distances). by adding this parameter, we can instead fallback to recovery behaviors and find a better path.
fix occasional boost::lock crash (#23)
limit the distance between poses as filter runs (#21)
improve the orientation filter, add tests (#20)
Contributors: Michael Ferguson

add usage documentation (#19) also remove unused min_vel_theta parameter
implement changes from review (#18) * Add collision checking during final rotation * Add comments about getRobotVel returning velocities * Rename path to simulated_path for clarity * Rename pose to target_pose and goal_pose (now separate variables) * Get rid of magic number for when to use final rotation
simulate intial rotation properly (#17) previously, this simulated the big arc that we wouldn't follow anyways. this could cause the robot to get stuck in corners or other tight areas.
Contributors: Michael Ferguson

add feature for final rotation (#15) if the final pose orientation is very different from the end of the path, we get a big sweeping turn. this feature uses the previous orientation to avoid that sweeping turn and instead does a final in-place rotation. Add filter and pose publisher.
not sure how build is working without this being executable
Contributors: Michael Ferguson

unitialized limits causes test failures (#14) limits does not initialize prune_plan to a value, causes flaky test when it ends up true (would also probably be bad on a real robot)
Contributors: Michael Ferguson