Photo Story

Photos: World’s first autonomous tandem drift performed in perfect sync

Drifting, a controlled skid, can teach vital recovery skills for improving road safety.

Published: Jul 31, 2024 11:16 AM EST

Mrigakshi Dixit

a month ago

Autonomous tandem drift sequence was performed at Thunderhill Raceway Park in California.

Toyota Research Institute

Every year, thousands of fatalities and accidents occur worldwide as a result of rapid loss of vehicle control. In such cases, drivers often have to act fast, and in a controlled manner to protect others on the road.

Autonomous technology may be the key to enhanced safety.

Now, Toyota Research Institute (TRI) and Stanford University have achieved the impossible: “the world’s first autonomous tandem drift sequence.”

Remarkably, the AI-powered self-driving cars aced controlled sideways moves while separated by barely inches.

Through this, Toyota and Stanford have pushed the boundaries of vehicle control by mastering the art of drifting. Drifting, a controlled skid, can teach vital recovery skills for improving road safety.

“The physics of drifting are actually similar to what a car might experience on snow or ice,” said Chris Gerdes, professor of mechanical engineering and co-director of the Center for Automotive Research at Stanford (CARS).

Gerdes added: “What we have learned from this autonomous drifting project has already led to new techniques for controlling automated vehicles safely on ice.”

1/7
Involved lead and chase cars
The experiments entailed two self-driving cars capable of drifting, which is analogous to operating a car on ice roads. The lead car and a chase car executed a synchronized drift, maneuvering within inches of each other at the limits of control. By incorporating a second car drifting in tandem, the researchers created a more realistic environment to test how vehicles respond to dynamic situations involving pedestrians, cyclists, and other cars.
2/7
Modified GR Supras
Two modified sports cars (Toyota's GR Supras) were used for the experiments at Thunderhill Raceway Park in California. The cars were equipped with AI systems, including a neural network tire model, to learn and adapt like a human driver. “The track conditions can change dramatically over a few minutes when the sun goes down. The AI we developed for this project learns from every trip we have taken to the track to handle this variation,” said Gerdes.
3/7
Cars equipped with AI systems
TRI developed the lead car's algorithms, while Stanford focused on the chase car. The lead car's control system was specialized for consistent and safe performance, while the chase car’s AI-enabled it to adapt to the leader's changing position and avoid collisions.
4/7
Computers and sensors for control
GReddy and Toyota Racing Development (TRD) customized the cars' suspension, engine, transmission, and safety features for racing. Both vehicles were built to Formula Drift standards for data collection with expert drivers. Moreover, the cars were outfitted with computers and sensors to control steering, acceleration, braking, and track their movement. A dedicated Wi-Fi network enabled real-time communication about their positions and intended paths.
5/7
Use of NMPC
The cars are required to constantly adjust their controls to drift together. Autonomous drifting requires constant adjustments to steering, throttle, and brakes. They use a system called Nonlinear Model Predictive Control (NMPC), which sets specific rules for how the cars should behave. “In NMPC, each vehicle starts with objectives, represented mathematically as rules or constraints that it must obey,” the press release noted.
6/7
Rapid control adjustments
The lead car was tasked with executing and sustaining a drift along a predetermined path within physical and vehicle limitations. The chase car's objective was to mirror the lead car's drift while implementing preventative measures to avoid a collision. As per the press release, the vehicles continuously adjusted their controls 50 times a second to optimize performance. “By leveraging AI to constantly train the neural network using data from previous tests, the vehicles improve from every trip to the track,” it noted.
7/7
Autonomous tech to prevent car accidents
Over 40,000 annual US fatalities and 1.35 million globally are linked to vehicle loss of control in sudden situations. Autonomous technology promises to prevent these. “When your car begins to skid or slide, you rely solely on your driving skills to avoid colliding with another vehicle, tree, or obstacle. An average driver struggles to manage these extreme circumstances, and a split second can mean the difference between life and death,” explained Avinash Balachandran, vice president of TRI’s Human Interactive Driving division. Balachandran further added: “This new technology can kick in precisely in time to safeguard a driver and manage a loss of control, just as an expert drifter would.”