HomeRobotics & KinematicsRuntime Safety Monitor 2D: Anomaly Detection & Fallback Switch

Runtime Safety Monitor 2D: Anomaly Detection & Fallback Switch

Interactive 2D top-down runtime-verification monitor for a learning robot: watch an EWMA divergence score between a corrupted policy and a verified-safe reference trajectory, tune the detection threshold and debounce window, and see when the monitor trips a hard switch to a safe fallback controller.

Robotics & Kinematics2DAdvanced60 FPS📱 Mobile-adapted⇄ 3D version
2d-robotics-topic-12 ↗ Open standalone

A wheeled robot patrols between two docks past a restricted zone using a potential-field policy whose hazard-avoidance gain degrades as you raise the perception-drift slider — modelling a sensor or perception fault rather than a hard-coded bug. An independent runtime monitor never inspects the policy itself; it only tracks how far the robot's actual trajectory diverges from a verified-safe reference ("ghost") running in parallel, folding that divergence into an EWMA anomaly score. Cross the tunable trip threshold for a debounce window of consecutive frames and the monitor hands control to the reference policy's own commands until the score settles back down — the same detect-and-switch pattern used by runtime-verification shields and supervisory watchdogs guarding learned robot controllers in the field. This top-down 2D rendition draws the exact same planar potential-field dynamics as the 3D version — the physics here was never anything but two-dimensional (x/z position, no vertical motion) — plus a live strip-chart of the anomaly score so you can watch the trip/release cycle unfold over time.

⚙ Under the hood

2D top-down runtime-verification monitor: watch a degraded robot policy drift toward a restricted zone while an independent monitor tracks its EWMA divergence from a verified-safe reference trajectory, tripping a hard fallback switch once the tunable threshold and debounce window are exceeded, alongside a live score-vs-threshold strip chart. Drag the top-down view to pan and scroll to zoom.

roboticsruntime verificationsafety monitoranomaly detectionfallback controlreinforcement learning

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install

What did you find?

Add reproduction steps (optional)