Neotix Robotics Data Platform

Full-stack web app for browsing, QA-reviewing, merging, and visualizing LeRobot v2.1 teleoperation datasets (YAM Pro single-arm & bimanual), with a live MuJoCo simulator, async sim-to-MP4 replay, and optional Tauri desktop build.

---

Overview

Operators manage robotics data through this platform instead of ad-hoc scripts: JWT auth, per-user datasets, multi-camera episode playback, QA workflows (validate / flag / delete, task edits, export), dataset merge with compatibility checks, activity audit log, and WebSocket-streamed interactive simulation. Sample datasets are pulled from Hugging Face; MuJoCo models live under models/i2rt_yam/.

---

Tech stack

Layer	Technologies
Backend	FastAPI, Uvicorn, SQLAlchemy, SQLite, JWT (python-jose), passlib/bcrypt
Data / sim	pandas, pyarrow, NumPy, MuJoCo, OpenCV (headless), Hugging Face Hub
Frontend	Vite 6, React 18, TypeScript, React Router 6, TanStack Query, Zustand
UI	Tailwind CSS v4, Radix UI, shadcn-style components, Framer Motion, Sonner
3D	Three.js, React Three Fiber, Drei, postprocessing
Charts	Visx, D3
Desktop (optional)	Tauri 2
Ops	Docker / Docker Compose, nginx (frontend image)

API docs: http://localhost:8000/docs (Swagger with Authorize for JWT).

---

Demo

---

Screenshots

Landing, auth, and marketing sections in a compact 3×2 grid:

Landing	Examples	Features
Workflow	Sign in	Sign up

---

Quick start

# Backend
python -m venv venv
venv\Scripts\activate          # Windows
pip install -r requirements.txt
uvicorn api.main:app --reload --host 0.0.0.0 --port 8000

# Frontend (new terminal)
cd frontend
npm install
npm run dev

Open http://localhost:3000 (or the port Vite prints). For video/API features, keep the backend on port 8000.

Optional — Tauri desktop: cd frontend && npm run tauri dev (Rust toolchain required). Ensure the dev server URL in frontend/src-tauri/tauri.conf.json matches Vite’s URL if the port is not 3000.

Optional — Docker: docker compose up --build (see docker-compose.yml).

Sample data: use huggingface-cli download as described in SUBMISSION.md.

---

Repository layout

• api/ — FastAPI app (routers: auth, datasets, episodes, qa, merge, activity, stats, simulator)
• frontend/ — Vite + React SPA; frontend/src-tauri/ — Tauri shell
• tools/ — Dataset loaders, merge utilities, sim replay helpers
• models/i2rt_yam/ — MuJoCo scenes (single + bimanual)

---

More detail

Implementation notes, checklist, design decisions, and full run instructions: SUBMISSION.md.