Nav2 Parameter Helper

Upload your map (.yaml + .pgm/.png) or an existing param file, set your robot's footprint, and tune AMCL & costmap parameters — the inflation layer's effect updates live on the map as you adjust it.

Map

Drop your map's .yaml and image (.pgm/.png) here, or browse

Already have an AMCL / nav2_params.yaml? Upload it to prefill the parameters below.

Robot footprint

Radius r (m)

This sets Nav2's robot_radius (circular) or footprint (rectangular, centered on the robot's origin) costmap parameter, and the derived inscribed radius used by the inflation layer's cost formula — for a rectangle that's half the shorter side.

AMCL parameters

min_particlesmax_particles

pf_err

pf_z

resample_interval

recovery_alpha_slow

recovery_alpha_fast

robot_model_type

alpha1

alpha2

alpha3

alpha4

alpha5

update_min_d (m)

update_min_a (rad)

transform_tolerance (s)

save_pose_rate (Hz)

tf_broadcast

Costmap parameters

Inflation layershow on map

inflation_radius (m)

cost_scaling_factor

inflate_unknowninflate_around_unknownenabled

Drag inflation_radius or cost_scaling_factor and watch the halo around obstacles on the map update live.

Export

Generates an amcl / local_costmap / global_costmap block in the same shape as nav2_params.yaml — paste it into your own file, or merge with the rest of your Nav2 stack config (planner, controller, behavior server, etc., which this tool doesn't generate).

Sample map (upload your own on the left) — click the map to move the footprint test point

Base map

free occupied unknown (unexplored) robot footprint (test point)

Inflation cost (overlaid on the map)

0 — no effect on planning253 — inscribed254 — lethal

Cost rises the closer a cell is to an obstacle. Lethal (254, solid red) means the robot's reference point being here guarantees a collision. Inscribed (253, red-orange) means the robot's own body would touch the obstacle even though its reference point hasn't reached it yet — this band's width is set by the footprint size. Below that, the blue→yellow gradient is cost decaying with distance out to inflation_radius; beyond it, cost is 0 and the layer has no effect.