TLDR;
Tom Shaw explores how AI can improve driving performance using a Formula E simulator at the Google Cloud Summit. The AI analyzes driving data, provides personalized coaching, and helps reduce lap times. The technology has broader applications in safety training, operating heavy machinery, and enhancing performance in various high-precision tasks.
- AI-driven analysis and coaching can significantly improve driving performance.
- The technology has potential applications beyond racing, including safety training and machinery operation.
- Real-time feedback and personalized coaching enhance learning and safety.
Intro – Can AI make me faster? [0:00]
Tom Shaw was invited to the Google Cloud Summit in London to test whether AI could help drivers improve their speed. He will use a Formula E simulator to gather driving data, which will then be analyzed by AI to provide insights and coaching. The project aims to demonstrate how AI can enhance driving performance.
Simulator setup & controls [0:24]
Before starting, Jack explains that Tom will complete three laps on the Formula E simulator. The simulator captures data from various sensors, which is sent to Google Cloud for driver analysis. The goal is to compare Tom's performance against a professional driver and use the AI-driven coaching to improve his lap times. The simulator setup involves specific controls: left foot for braking, right foot for accelerating, and standard steering.
First three‑lap run [1:12]
Tom begins his first three-lap run on the simulator. He experiences some difficulties, particularly with braking too deeply in the corners. Despite these challenges, he completes the laps, aiming to improve in the subsequent runs based on the data collected.
AI analysis of my driving [2:28]
After the first run, the AI analyzes Tom's driving data using Gemini, comparing it to a professional driver's performance. The analysis reveals that Tom's initial launch speed was slightly slower and that he braked too harshly, lacking smoothness through the corners. The key takeaway is that Tom needs to be smoother and more relaxed in his driving style to improve his lap times.
How the data pipeline works [3:26]
Dan from Formula E explains the data pipeline behind the AI analysis. The driving data from the rFactor 2 software is collected via an agent on the simulator and sent to a BigQuery database. This data is then compared to both professional driver data and theoretical "perfect laps" to provide tailored feedback. The large language model tailors the feedback to the driver's skill level, offering relevant and actionable advice.
Real‑time coaching plans [5:17]
Google Cloud aims to evolve this technology from "race to roads," applying insights from the racetrack to real-world driving scenarios. Currently, feedback is provided in written form, but the plan is to develop real-time coaching through earbuds, offering constant feedback on driving performance. This real-time feedback can help drivers maintain safety, correct bad habits, and improve overall driving skills, with applications in workforce safety and education.
Second run on the sim [6:01]
Fully informed by the AI feedback and Dan's insights, Tom undertakes a second three-lap run on the simulator. His goal is to apply the coaching and achieve a faster lap time than his initial attempt.
Lap‑time results (Was I faster?) [6:43]
The results show that Tom significantly improved his lap time on the second run. His fastest time in the first run was 1 minute 10.728 seconds, while his fastest time in the second run was 1 minute 8.201 seconds, making him over two seconds faster. This demonstrates the effectiveness of the AI-driven coaching in enhancing driving performance.
Real‑world applications beyond racing [7:06]
The technology has numerous real-world applications beyond motorsport. It can be used to train individuals operating heavy machinery, ensuring they adhere to safety guidelines. Additionally, it can assist new drivers in learning to drive and improve performance in any high-precision task. The potential use cases are vast and could transform how people interact with objects, machinery, and tools.
Outro & thanks [7:24]
Tom thanks Google Cloud for sponsoring the video and hosting him at the event. He encourages viewers to explore the Google Cloud blog for more information about the AI agent used in the project.