Loading system evidence
Loading system evidence
Diagnose navigation failures by tracing lifecycle, costmaps, planners, controllers, behavior trees, and mission state.
Sign in to start - freeOutcome
Build
A resilient delivery-and-docking mission with measurable recovery behavior.
Sample incident
Activate the required server without restarting the simulation.
Full syllabus
Trace a navigation request through navigator, behavior tree, planner, controller, costmaps, and lifecycle state.
Trace sensor observations through marking, clearing, range, persistence, and frame configuration.
Relate physical clearance to footprint radius, inflation distance, cost scaling, and map resolution.
Inspect planner inputs, unknown space, resolution, tolerances, and search output.
Interpret velocity constraints, trajectory critics, progress checks, and command output.
Trace node status, blackboard values, retries, timeouts, and recovery selection.
Separate completion criteria, progress monitoring, oscillation detection, and protective stop behavior.
Coordinate task queues, docking, charging, retries, and operator intervention around Nav2.
Two localization bugs pulled straight from a working warehouse robot: a guide camera mounted flipped, and a pose estimate that teleports on a bad fix.
Two controller bugs from the warehouse robot: it overshoots the dock, and it jitters the wheels on every plan update.
Playgrounds
Small, functional experiments that reveal the math and evidence behind this course’s failures. Open them alongside the labs.
Tune robot footprint, map resolution, inflation radius, and scaling to understand blocked paths.
costmap inflation / Nav2 / robot footprint
Live controlsWatch a particle filter converge (or not) as you change particle count, initial pose error, sensor noise, and the loaded map.
AMCL / particle filter / localization
Live controlsSeparate 'is there a valid plan' from 'can the controller follow it' by tuning goal tolerance, lookahead, corridor width, and footprint.
Nav2 planner / controller / goal tolerance