GRaceMind

Inspiration Modern racing is no longer just horsepower and reflexes—it’s a data war. Every lap generates thousands of data points, but extracting real-time insight requires engineering tools that only top-tier professional teams usually have. We wanted to democratize that. The inspiration behind GRaceMind was simple: What if any driver, crew chief, or engineer could get strategy-grade intelligence instantly—powered by Toyota GR telemetry, AI, and predictive modeling? Our goal became clear: Turn raw GR Cup data into a real-time racing mind. 🧠 What it does GRaceMind is a real-time race strategy and driver insights engine built on Toyota GR telemetry. It transforms data into actionable intelligence through:

1. Live Telemetry Reconstruction Simulates real-time data streaming from historical GR Cup datasets.
2. Predictive Pace Modeling Forecasts lap times using driver inputs, tire state, track sectors, and vehicle dynamics.
3. Tire Degradation Estimation AI-driven wear curve that predicts when performance drop-off becomes critical.
4. Pit Window Optimization Runs multiple strategy simulations to recommend the optimal pit lap—minimizing traffic, maximizing track position.
5. Driver Performance Insights Identifies opportunities for improvement: braking points throttle modulation corner exit consistency sector-specific weaknesses
6. Real-Time Alerts Detects anomalies such as pace drops, overheating patterns, or inconsistent lines.
7. Interactive Dashboard A simple interface showing: live telemetry track map positioning predicted lap time tire wear curve strategy recommendation lap delta chart GRaceMind is designed to feel like a race engineer sitting beside you, updating every second. 🏗️ How we built it We developed GRaceMind in a fully modular architecture: Data Layer Parsed TRD hackathon datasets Built a replay server that emits streaming JSON packets at 5–10 Hz Standardized telemetry across speed, RPM, throttle, brake, gear, GPS, lap, and sector Feature Engineering Created a pipeline that extracts: sector splits micro-events pace deltas tire wear proxies rolling window stats track-map projections via Shapely Machine Learning Models Baseline: LightGBM regression for lap-time prediction Advanced: LSTM for sequential driver-performance modeling Tire model: Rolling degradation curve + regression Pit optimizer: Heuristic + Monte Carlo simulation Backend (FastAPI) WebSocket telemetry stream Predictive inference endpoint Unified strategy engine Frontend (Streamlit) Live telemetry gauges Real-time lap forecasts Tire wear visualization Strategy & pit recommendations Driver performance analytics Deployment A full Docker Compose setup spins up: backend models dashboard replay server One command: docker-compose up 🔧 Challenges we ran into Telemetry cleaning — GPS drift, noisy RPM samples, inconsistent timing stamps Track reconstruction — converting raw GPS into a normalized XY map for visualization Predicting pace — driver variability, tire wear, and temperature all influence lap deltas Real-time streaming — ensuring sub-200ms latency across inference + UI Pit strategy modeling — balancing simplicity for MVP with realistic behavior 🏆 Accomplishments we’re proud of Built a fully functional real-time analytics engine from scratch Achieved accurate baseline lap time prediction using GR Cup data Delivered a live dashboard that feels like a true motorsport engineering tool Created a reusable, extensible racing analytics framework Packaged everything into one-click Docker deployment Designed GRaceMind to scale into a full digital twin of race sessions 📚 What we learned Motorsport telemetry is incredibly high-dimensional—and requires domain understanding Driver behavior has consistent signatures that models can learn Real-time predictive systems must balance accuracy with latency Strategy is fundamentally probabilistic: small deltas compound into race-changing decisions Visualizing racing data is as important as modeling it GRaceMind taught us how to merge AI, data engineering, motorsport physics, time series modeling, and UX design into one cohesive, high-performance system. 🚀 What’s next for GRaceMind We’re just getting started. Here’s what’s coming next:
8. Full Digital Race Twin Replay any GR Cup race with complete simulations of driver behavior and tire dynamics.
9. Driver Coaching Mode Post-session analysis that automatically generates personalized improvement plans.
10. Advanced AI Models Transformer-based pace estimator Graph neural networks for track topology Real-time multi-driver strategy engine
11. Cloud-Connected Telemetry Ingest live WiFi/5G data from cars during actual events.
12. Team Collaboration Tools Crew dashboard Race engineer cockpit Driver post-race debrief generator
13. GRaceMind Mobile App Pocket-sized live race intelligence for drivers and fans.

Built With

• all