Toyota Research Institute and Stanford Engineering have achieved a groundbreaking milestone in autonomous driving by successfully completing the world’s first autonomous tandem drift with two GR Supra vehicles. This pioneering achievement aims to enhance vehicle safety and responsiveness by simulating conditions where cars must react swiftly to external factors. For nearly seven years, the two organizations have collaborated on making driving safer by focusing on the principles of drifting.
The technique helps simulate conditions where cars must react quickly to other vehicles, pedestrians, or cyclists. Avinash Balachandran, vice president of TRI’s Human Interactive Driving division, said their researchers came together with the goal of making driving safer. He noted that this milestone demonstrates they can control cars dynamically at the extremes, laying the foundation for advanced automotive safety.
Chris Gerdes, a professor of mechanical engineering and co-director of Stanford’s Center for Automotive Research, added that the physics of drifting are similar to what a car experiences on snow or ice. He said their findings from this project have already led to new techniques for controlling automated vehicles safely in such conditions. In the autonomous tandem drift sequence, two vehicles—a lead car and a chase car—navigate a track in close proximity, operating at the edge of control.
Ai advances in autonomous drifting
Using advanced AI techniques, including a neural network tire model, the cars adapt to track conditions in real-time. The experiment was conducted at Thunderhill Raceway Park in Willows, California, using two modified GR Supras.
TRI focused on the lead car, developing stable control mechanisms, while Stanford tackled the chase car, ensuring it could drift without colliding. Both cars were modified by GReddy and Toyota Racing Development to meet Formula Drift specifications, including enhancements to suspension, engine, transmission, and safety systems like roll cages and fire suppression. The vehicles are equipped with computers and sensors that allow for real-time control and communication via a dedicated WiFi network.
To achieve autonomous tandem drifting, the vehicles continually plan and adjust their steering, throttle, and brake commands using a Nonlinear Model Predictive Control technique. This involves solving an optimization problem up to 50 times per second, allowing the vehicles to react to rapidly changing conditions. AI continually trains the neural network using data from previous tests, enhancing the vehicles’ performance with each run.
The innovative work done by TRI and Stanford Engineering represents a promising leap towards the future, where cars can operate more safely under extreme conditions.