A new open-source project called Steam Controller Auto-Charge automatically docks Steam Controllers using computer vision and haptic feedback. The system uses an overhead camera with OpenCV.js to track the controller and charging puck, then guides the device into place through 70Hz haptic pulses sent via Web HID to the controller's dual Linear Resonant Actuators.

The project combines Lucas-Kanade optical flow algorithms with a Rust/WASM-powered neural network for object detection, running entirely in the browser. When the controller nears the puck within 150 pixels, it switches to a gentler 50% reduced haptic frequency for precise magnetic docking. Battery status polling monitors charging success through Report ID 121 and displays live battery percentage via Report ID 67 telemetry.

Setup requires only the Nix package manager and a Chromium-based browser supporting Web HID. Users mount an overhead webcam, place the puck and controller on desk, then launch with a single nix-shell command. The system automatically fetches dependencies and builds the WASM module for real-time visual processing.

This represents a novel application of browser-based robotics, demonstrating how modern web APIs can enable sophisticated hardware control without native drivers. The project proves that complex computer vision tasks can run efficiently in-browser while maintaining hardware-level precision.