Inspiration
The thrill of Formula 1 racing and the critical role of real-time data in split-second decisions inspired this project. Watching pit crews make strategic calls based on telemetry data, we realized how challenging it must be to maintain reliable data transfer in high-speed, unstable network environments. The problem statement's focus on "racetrack telemetry" perfectly aligned with our passion for motorsports and IoT technology.
What it does
RaceEdge simulates a complete F1 telemetry ecosystem using consumer hardware. An ESP32 acts as the race car, collecting sensor data and transmitting it wirelessly to a pit station laptop that performs real-time ML-powered data quality control. The cleaned data flows to a factory system that uses advanced ML models for race strategy optimization and performance analysis, creating actionable insights for competitive advantage.
How we built it
We designed a three-tier architecture: ESP32 for data collection, pit station for real-time ML processing, and factory for advanced analytics. The ESP32 handles sensor integration and WiFi communication. The pit station runs Python-based ML models using scikit-learn for anomaly detection and data cleaning. The factory system leverages TensorFlow and XGBoost for strategy optimization and performance prediction. All components communicate via HTTP/WebSocket protocols with JSON data formatting.
Challenges we ran into
Network reliability: Simulating unstable racing conditions while maintaining real-time data flow ML model performance: Balancing accuracy with low-latency requirements for pit station processing Data synchronization: Coordinating three different systems with varying processing speeds Resource constraints: Implementing meaningful ML on ESP32's limited computational power
Accomplishments that we're proud of
Created a realistic F1 data pipeline that addresses real-world telemetry challenges Successfully integrated ML at both pit and factory levels with measurable performance metrics Achieved sub-100ms processing latency for critical data quality control Designed a scalable architecture that could be adapted for actual racing applications
What we learned
Real-time ML requires careful balance between model complexity and processing speed Network resilience is crucial for mission-critical telemetry systems Data quality preprocessing significantly improves downstream ML model performance IoT systems benefit from hierarchical processing (edge → cloud) architecture
What's next for Race Edge
Implement edge ML on ESP32 for local anomaly detection Add predictive maintenance capabilities using vibration analysis Integrate computer vision for driver behavior analysis Expand to support multiple car simulation for team-wide strategy optimization Deploy on actual go-kart for real-world testing and validation
Log in or sign up for Devpost to join the conversation.