Tuning move_base to avoid in place rotation

asked 2020-04-04 05:43:46 -0500

Scarf195 gravatar image

Hi, I've implemented Navigation Stack for autonomous navigation of a differential-wheel robot. In particolar, my vehicle is composed by two driving wheels and a caster wheel. Becouse of the gearmotor the vehicle can't rotate in place for low rotational velocities (example 0.5 rad/sec) and would not set the parameter min_in_place_vel_theta to an high value becouse this will cause lot of problems during navigation.

So is possible to tuning planner parameter in order to avoid in-place rotation?

This is for now the dwa_local_planner_params.yaml file:

DWAPlannerROS:

max_vel_x: 0.2

min_vel_x: 0.10

max_vel_y: 0.0

min_vel_y: 0.0

max_vel_trans: 0.2

min_vel_trans: 0.001

max_vel_theta: 1.5

min_vel_theta: 0.8

min_in_place_vel_theta: 1.0

acc_lim_x: 2.0

acc_lim_y: 0.0

acc_lim_theta: 8.0

xy_goal_tolerance: 0.05

yaw_goal_tolerance: 0.17

latch_xy_goal_tolerance: false

sim_time: 1.5

vx_samples: 10

vy_samples: 0

vth_samples: 20

controller_frequency: 2.0

path_distance_bias: 32.0

goal_distance_bias: 20.0

occdist_scale: 0.01

forward_point_distance: 0.1

stop_time_buffer: 0.2

scaling_speed: 0.25

max_scaling_factor: 0.2

oscillation_reset_dist: 0.05

publish_traj_pc : true

publish_cost_grid_pc: true

edit retag flag offensive close merge delete

Comments

1

How about switching to another local planner, e.g. teb_local_planner which is more suitable for car-like motion.

Humpelstilzchen gravatar image Humpelstilzchen  ( 2020-04-05 01:56:10 -0500 )edit

Hi, thank you for your reply! Yes I've tried also teb_local_planner but it requires too much computational effort (I'm using rtab_map and Navigation Stack on Jetson Nano). Can I get any results by playing with the scoring parameters?

Scarf195 gravatar image Scarf195  ( 2020-04-05 10:42:01 -0500 )edit