Race Engineer Dashboard

Team Members

Donte & Adi

---

---

Summary

Live telemetry and analysis for the car: sensor streams (with filtering), derived run metrics, and optional RaceGPT insights in a sidebar.

Sensor views

• Speed time series, live, and max value
• Power time series (calculated from current and voltage data)
• GPS location display with Google Maps
• Live Steering angle, Brake pressure, Throttle, and RPM on all wheels
• Timestamps and stopwatch for lap and race timing

Derived metrics

• Distance calculated from aggregating GPS data
• Energy Use calculated from power
• Efficiency instantaneous and average over a run

RaceGPT

• Configurable manual or automatic LLM requests that analyze recent telemetry data and return verdicts on improving performance

---

Getting Started

1. Running the Project

   Create a .env file in the the root directory (not under /backend or /frontend). See .env.example for more info.

   Make sure Docker containers and volumes for this project are not running already. Then, run docker compose up --build to get all containers running.

   The frontend client UI should be running on port 3000.
   The backend healthcheck endpoint is on the root of port 8000.

   Integration with RaceGPT is done over a websocket connection to a machine running RaceGPT. Connect your machine to the RaceGPT host, and the dashboard should connect when built.

2. Frontend Testing

   Refer to frontend/README.md for more information.
   Bun Installation: The frontend uses Bun instead of NodeJS as a package manager and runtime. The installation is linked here. Then, run the following commands in the terminal to test/run the frontend in isolation.
   This will start the frontend development environment with HMR and Vite at port 5173.

   cd frontend
   bun dev

   Google Maps: To get location data and Google Maps properly displaying while running only the frontend with bun dev, create a .env file in the /frontend directory. Follow the .env.example in the /frontend and create the environment variables below.

   VITE_GOOGLE_MAPS_API_KEY=<Your Google Maps API Key here>
   VITE_GOOGLE_MAP_ID=<Your Google Maps Map ID from Google Cloud console>

   Instructions for getting your own API_KEY and MAP_ID are in frontend/README.md.

3. Backend Troubleshooting

   Refer to backend/README.md.

---

ROS Subscriber Data

For the ROS2 subscriber, first get the ROS2 publisher IP address from:

ssh cev@<daq tailscale ip> "docker exec ts-authkey-container tailscale ip" | head -n1

Make sure this matches the ip in the docker-compose.yml file.

We set it manually, not via env var, because the publisher IP should not change.

RaceGPT Integration

Connect your machine to another machine running RaceGPT so the dashboard can reach the RaceGPT websocket service and request analysis on live telemetry snapshots.

As mentioned above, there are two modes for requesting responses on the sidebar.

• Manually request LLM responses (5s buffer between responses)
• Automatically request and display LLM responses based on a set frequency >5s

---

System Overview

ROS2 Sensors → Backend (Python + ROS2) → WebSocket Stream → Frontend (React + TypeScript + Bun)
                                          ↓
                                  RaceGPT Integration

---

RaceGPT Integration Flow

RaceGPT is integrated as an on-demand analysis layer.

1. The frontend sends telemetry history to the backend via a POST request
2. The backend forwards this data over a USB WebSocket connection to the RaceGPT machine (/ws/analyze)
3. RaceGPT processes the data and returns a verdict/analysis
4. The backend relays the response back to the frontend
5. The frontend displays the result in the sidebar (manual or automatic modes)

---

ROSbag + Remote Data Handling

• The frontend can trigger ROSbag recording via /bag endpoints
• The backend communicates with the DAQ machine through Tailscale
• ROSbag files are stored remotely for later analysis and replay