robotics-perception-planning

name	robotics-perception-planning
description	Design and debug robotics pipelines spanning perception, localization, mapping, planning, control interfaces, and sim-to-real transfer. Use when evaluating a robot stack, diagnosing failures in closed-loop behavior, planning experiments, or comparing perception-driven versus policy-driven system designs.

Robotics Perception Planning

Use this skill when robot behavior emerges from multiple coupled modules and a single metric hides the real failure.

Core Workflow

1. Define the robot task, sensing stack, actuation limits, and success criteria.
2. Map the closed loop:
   • sensing,
   • state estimation,
   • world model,
   • planning,
   • control interface,
   • safety interlocks.
3. Use references/sim2real-checklist.md when simulation results do not transfer.
4. Separate open-loop module quality from closed-loop task success.
5. End with the next experiment that most cleanly isolates the failure source.

Execution Rules

1. Track latency and synchronization between sensors and control decisions.
2. Treat calibration, frame transforms, and state estimation drift as first-class suspects.
3. Separate planner weakness from perception error and actuator saturation.
4. Keep safety and fallback behavior explicit.

Output Contract

Return:

1. Closed-loop map.
2. Failure decomposition.
3. Measurement plan.
4. Sim-to-real risks.
5. Best next experiment.