🤖 Robotic Harness

An embodied-intelligence research plugin suite for DeepSeek Harness

Put robot assets, simulation, capability orchestration and failure evidence into one Agent workflow — from CAD/URDF inspection to MuJoCo pick-and-place, fault injection, evidence-based diagnostics and reproducible experiment bundles.

🧪 Testers & contributors: this project is at an early Demo stage. It has been validated only in a limited local environment (Windows + Anaconda Python 3.10 + DSH 0.1.0-rc.6). ROS 2, CAD, real-robot and other OS/hardware setups have not been fully tested — please forgive rough edges and report any issues you hit. Everyone is welcome to test, modify and extend this plugin suite, and to combine your own robot-related plugins (ROS 2 / CAD / vision / control / VLA / ...) with this suite to build one bigger, complete robot plugin bundle together — every independent module can be published and contributed separately. See CONTRIBUTING.md.

---

📑 Table of contents

• ✨ Features
• 📸 Screenshots
• 🚀 Quick start (30 seconds)
• 🏗️ Architecture
• 📦 Install as a DSH plugin
• 🧩 Tool & skill surface
• 🎯 The demo (MuJoCo pick-and-place)
• ⚠️ Known limitations
• 🌊 Future vision
• 📂 Repository layout
• 📚 Documentation
• 🤝 Contributing
• 📄 License

---

✨ Features

🔍 Assets & CAD	URDF / MJCF / SDF inspection, inertia & topology validation, mesh stats, SVG preview, URDF→MJCF conversion, SDF-compat export, CAD inventory & version compare
🎮 Simulation	MuJoCo pick-and-place with 6 fault-injection modes, batch benchmarks, read-only replay, sim-vs-real gap reports
🧮 Control	Tracking metrics (rise/settle/overshoot/SSE), trajectory validation, planned-vs-actual compare, PID templates & config compare, system identification
👁️ Vision	Color/generic perception routing, camera health, calibration inspection, pose checks, perception comparison, failure-frame annotation
🧠 Embodied models	Model registry, builtin demo adapters (run for real), honest backend probes, rule-based capability routing, policy rollout compare
📡 Telemetry & diagnostics	Deterministic rule engine (facts / rules / hypotheses), anomaly scan, failure-evidence collection, run compare
🧬 Data pipeline	Inventory, schema, time-sync, alignment, non-destructive transforms, episodes, leakage-safe splits, de-identification, rosbag conversion, LeRobot export, dataset versions & cards
🔬 Experiments	Spec → matrix → benchmark → metrics → ablation → report
🤖 Real-robot flow	Preflight checklist + experiment state machine (hardware items skipped honestly without an adapter)
📚 Knowledge	Docs index/search, error-code lookup, diagnostic-case search, project memory (retrieve/ingest)
📖 Research	Literature search (arXiv / Semantic Scholar) + problem→solution proposals with evidence for any stage
🚀 Autonomous training	Server check → training plan → supplementary dataset discovery → job prepare (dry-run default) → confirmed remote submit → status → report
📊 **R

…