Rider-PC Client

PC-side client infrastructure for the Rider-PI robot, providing:

• REST API adapter for consuming Rider-PI endpoints
• ZMQ subscriber for real-time data streams
• Local SQLite cache for buffering data
• FastAPI web server replicating the Rider-PI UI
• AI Provider Layer with real ML models (Voice, Vision, Text)
• Production-ready deployment with Docker and CI/CD

🎉 Phase 4 Complete: Real AI Models & Production Deployment

This project now includes:

• ✅ Real AI Models: Whisper ASR, Piper TTS, YOLOv8 Vision, Ollama LLM
• ✅ Docker Deployment: Complete stack with Redis, Prometheus, Grafana
• ✅ CI/CD Pipeline: Automated testing, security scanning, Docker builds
• ✅ Health Probes: Kubernetes-ready liveness and readiness endpoints
• ✅ Automatic Fallback: Mock mode when models unavailable

Quick Start

Option 1: Docker (Recommended)

# Create .env file
echo "RIDER_PI_HOST=192.168.1.100" > .env

# Start the full stack
docker-compose up -d

# Access services
# Rider-PC UI: http://localhost:8000
# Prometheus: http://localhost:9090
# Grafana: http://localhost:3000

Option 2: Local Development

# Install dependencies
pip install -r requirements.txt

# Run in mock mode (no AI models required)
python -m pc_client.main

Development Workflow

• Python: develop with Python 3.11 (CI target) while keeping code compatible with Rider-PI’s Python 3.9.
• Tooling: install lightweight dev deps and hooks:

  pip install -r requirements-ci.txt
  pre-commit install

• Checks: use the Make targets to stay aligned with CI:
  • make lint → ruff check .
  • make format → ruff format .
  • make test → pytest suite with async/timeouts configured like GitHub Actions.
• CI split: every push to main goes through the Quick Checks workflow (ruff + short unit tests). The full pipeline (unit-tests, e2e-tests, css-ui-audit, Copilot setup) is run on pull requests.
• Copilot / agent flow: uruchom ./config/agent/run_tests.sh, aby odtworzyć środowisko wykorzystywane przez GitHub Copilot coding agent.
  • Skrypt instaluje zależności z config/agent/constraints.txt i odpala pytest.
  • Szczegółowa checklista PR-ów: docs/CONTRIBUTING.md.
• Rider-PI Integration:
  • UI (web/control.html) now mirrors Rider-Pi, including “AI Mode” and “Provider Control” cards.
  • Backend proxies /api/system/ai-mode and /api/providers/* to Rider-Pi via the RestAdapter, caching results for offline development.
  • To test locally with the real device, update .env (RIDER_PI_HOST, RIDER_PI_PORT) and run make start, then open http://localhost:8000/web/control.html.
  • Vision offload is now wired end-to-end: set ENABLE_PROVIDERS=true, ENABLE_TASK_QUEUE=true, ENABLE_VISION_OFFLOAD=true, and point TELEMETRY_ZMQ_HOST at Rider-Pi so the PC publishes vision.obstacle.enhanced after processing vision.frame.offload.
  • Voice offload mirrors the same flow: ENABLE_VOICE_OFFLOAD=true lets Rider-PC consume voice.asr.request / voice.tts.request, run Whisper/Piper (or mock), and publish voice.asr.result / voice.tts.chunk back to Rider-Pi for immediate playback.
  • Text/LLM integration exposes /providers/text/generate plus a capability handshake (GET /providers/capabilities) so Rider-PI knows which domains/versions Rider-PC supports before przełączeniem.

Tests & Maintenance

• Activate the virtualenv before checks: source .venv/bin/activate || true.
• Type check: mypy (expected: Success: no issues found in 75 source files).
• Quick test run: pytest -q.

Documentation

📚 Full Documentation - Complete documentation and guides

Localization & Translation Status

• English UI coverage: all /web/*.html screens now rely on web/assets/i18n.js keys, the Polish strings have English fallbacks, and the chat/chat-pc/control/system scripts no longer declare duplicate helpers.
• Verification: run node scripts/check_i18n.mjs to ensure every key includes an English value, and pytest (inside .venv) covers the back-end regressions tied to UI rendering.
• Notes: run ?lang=en against view, control, navigation, system, models, project, assistant, chat, chat-pc, google_home, mode, and providers to keep translations in sync.

Quick Links

• Quick Start Guide - Get started quickly
• AI Model Configuration - Setup real AI models
• Architecture - System architecture overview
• Configuration Hub - Central configuration guide
• API Documentation - REST API reference
• Replication Notes - Notes for replicating the project

Configuration Guides

• AI Model Configuration - Whisper, Piper, YOLOv8, Ollama setup
• Security Configuration - WireGuard VPN, mTLS setup
• Task Queue Configuration - Redis, RabbitMQ setup
• Monitoring Configuration - Prometheus, Grafana setup

Operations

• PC Offload Integration - Enabling AI mode / provider parity between Rider-Pi and Rider-PC
• Service Management - Operations, monitoring, troubleshooting
• Future Work - Planned improvements and development

See Also

• Rider-PI Repository

📝 License

Distributed under the MIT License. See LICENSE for more information.

Copyright (c) 2025-2026 Maciej Pieniak