How to prevent navigation path from hugging close to obstacles
Hello All, I have an omnidirectional robot and I am using DWA Local Planner on Noetic. I followed the fine tuning guide but my navigation path get very close to obstacles on the right when going between furniture to the point where it Aborts navigation or rotates to recover then aborts. My cosmap, global, DWA config and Rviz image is below. Any advice on how to fix this?
Costmap_Common
resolution: 0.05
robot_radius: 0.22
plugins:
- { name: static_map, type: "costmap_2d::StaticLayer"}
- { name: sensor, type: "costmap_2d::ObstacleLayer"}
- { name: inflation, type: "costmap_2d::InflationLayer"}
sensor:
observation_sources: laser_scan_sensor
laser_scan_sensor: {sensor_frame: base_laser, data_type: LaserScan, topic: scan, marking: true, clearing: true, track_unknown_space: true, observation_persistance: 0.0, obstacle_range: 2.5, raytrace_range: 6.0, inf_is_valid: true}
combination_method: 1
inflation:
inflation_radius: 0.75 #was 1.25
Global_Planner
base_global_planner: "global_planner/GlobalPlanner"
allow_unknown: false
use_dijkstra: true
use_quadratic: true
use_grid_path: false
old_navfn_behavoir: false
cost_factor: 0.55
neutral_cost: 66
lethal_cost: 253
#resolution: 0.05
cost_scaling_factor: 2.58 #was 2.58
DWA_Planner
base_local_planner: "dwa_local_planner/DWAPlannerROS"
controller_frequency: 5.0 # 10.0 could be a problem for control loop error
controller_patience: 15.0 # 15.0
DWAPlannerROS:
holonomic_robot: true
# Tolerances Defaults
yaw_goal_tolerance: 0.05 # was .05
xy_goal_tolerance: 0.10 # 0.10 **was0.05same above
latch_xy_goal_tolerance: false # false
prune_plan: true
# Forward simulation Defaults
use_dwa: true # true
sim_time: 3.5 # was 2.3 before**
sim_granularity: 0.025 # 0.025
angular_sim_granularity: 0.017 # 0.10 (1 degree = 0.017 rad)
vx_samples: 13 # 3 **was default
vy_samples: 13 # 10 **was default
vth_samples: 20 # 20 **was default
# Scoring Defaults
meter_scoring: false # Is this even taken into account?
path_distance_bias: 32.0 # 32.0**was1**B10
goal_distance_bias: 20.0 # 24.0 was 0.10 *****B32
occdist_scale: 0.05 # 0.02 ******B0.05
forward_point_distance: 0.325 # 0.325
stop_time_buffer: 0.20 # 0.20
scaling_speed: 0.25 # 0.25
max_scaling_factor: 0.20 # 0.20
heading_lookahead: 0.325
heading_scoring_timestep: 0.8
heading_scoring: true # Is this even taken into account?
penalize_negative_x: false # Is this even taken into account?
# Acceleration and velocity limits
# acc_limit_trans: 1.0
acc_lim_x: 1.0 # 2.5
acc_lim_y: 1.0 # 2.5
acc_lim_theta: 1.0 # 3.2
max_vel_x: +0.20
min_vel_x: -0.20 #was -0.22 #t
max_vel_y: +0.20
min_vel_y: -0.20
max_vel_theta: 0.2 #t
min_vel_theta: -0.2 #ch
# Other Defaults
oscillation_reset_dist: 0.05 # 0.05
#oscillation_reset_angle: 0.20 # 0.20
#prune_plan: false # true
#trans_stopped_vel: 0.1 # 0.1
theta_stopped_vel: 0.1 # 0.1
#publish_cost_grid_pc: true # false
C:\fakepath\ClosetoObstacle (2).png
Asked by chm007 on 2023-06-30 11:18:48 UTC
Comments