AI
Revolutionizing Safety: Toyota and Stanford’s AI-Powered Drift Cars Push the Limits of Autonomous Driving
To go back to this article, go to My Profile, and then check out the stories you've saved
Toyota Executes a Thrilling Demonstration Featuring Two AI-Controlled Race Cars Drifting in Unison
Typically, skidding during high-speed driving signals trouble. However, researchers from the Toyota Research Institute and Stanford University have engineered two autonomous vehicles that intelligently manage to skid or "drift" on purpose. This maneuver is designed to explore the extreme capabilities of self-driving technology.
In May, at the Thunderhill Raceway Park located in Willows, California, two self-driving cars executed a breathtaking maneuver by drifting in sync. This thrilling feat was showcased in a promotional clip where the vehicles, only inches apart, zoomed around the track, moments after the human drivers had stepped back, handing over the controls.
Stanford University's Chris Gerdes, who was at the forefront of the project, shared with WIRED that the methodologies crafted during this endeavor might pave the way for advancements in upcoming driver-assistance technologies. "We're exploring the potential to match or even surpass the capabilities of top-tier human drivers," says Gerdes.
Advanced assistance systems for drivers in the future could apply the algorithms trialed on the California circuit to step in when a driver is no longer in command, expertly maneuvering the vehicle to safety akin to the actions of a professional stunt driver. "The work we've accomplished can be expanded to address bigger challenges such as autonomous navigation in city environments," states Gerdes.
A pair of Toyota Supras performing a synchronized drift at the Toyota Research Institute. Photo credit: Toyota Research Institute
The initiative showcases an impressive example of rapid autonomous movement, yet autonomous cars have yet to reach flawlessness. Despite ten years filled with bold claims and excitement, driverless cabs are now functional in certain restricted scenarios. Nonetheless, these automobiles often encounter difficulties that leave them immobilized, necessitating support from afar.
Researchers from Toyota and Stanford University equipped two GR Supra sports cars with advanced computing hardware and sensory equipment to monitor the road, other vehicles, and the cars' suspension among other characteristics. Furthermore, they created algorithms that merge sophisticated mathematical representations of tire and track attributes with machine learning techniques, enabling the vehicles to autonomously learn and perfect the technique of drifting.
Ming Lin, an academic at the University of Maryland specializing in autonomous driving, describes the progress as a significant step forward in enabling self-driving vehicles to function under extreme conditions. "Navigating safely during adverse weather conditions such as rain, snow, or fog, or in dimly lit environments at night, remains one of the paramount hurdles for autonomous vehicles," she notes.
Lin emphasizes that the collaboration between Toyota and Stanford highlights the critical value of integrating machine learning with real-world physical models. “Even though it's still in the preliminary stages, it’s evidently moving towards the right path,” she notes.
In 2022, Toyota and Stanford showcased technology enabling self-driving cars to perform drifts. Executing this maneuver simultaneously with two vehicles demands superior precision and necessitates inter-vehicle communication. Professional drivers' lap data served as input for the cars, which then, through their onboard computers, solved an optimization problem as frequently as 50 times per second to adjust steering, acceleration, and braking accordingly.
"In this scenario, the focus is on mastering vehicle control under extreme performance situations, such as when the tires lose grip, akin to conditions experienced while driving on snow or ice," explains Avinash Balachandran, the vice president of the Human Interactive Driving division at TRI. "Regarding safety, merely being a competent driver doesn't cut it, which is why we're aiming to gain insights from the most skilled professionals."
Recently, the globe has witnessed significant progress in artificial intelligence, primarily due to the expansive language models that fuel applications such as ChatGPT. Yet, as evidenced by the dual drifting demonstration, successfully navigating the chaotic and unforeseeable realm of the physical world continues to be a distinct challenge.
"Balachandran notes that when a large language model makes a factual error, often referred to as a hallucination, it might not be a catastrophic issue. However, he points out that the stakes could be significantly higher in the context of an automobile."
In Other News…
Exclusive: An email informed Joe Biden's campaign staff that he was dropping out of the race, instructing them to look at X
J.D. Vance's Venmo account was left visible to the public, revealing its contents.
Delivered to your email: Receive Plaintext—Steven Levy's in-depth perspective on the worldwide influence of technology
Apple, Nvidia, and Anthropic utilized thousands of YouTube clips without permission to educate artificial intelligence systems.
Purchasing generic versions of Ozempic online is surprisingly simple, regardless of whether you actually require it or not.
The WIRED AI Elections Initiative: Monitoring over 60 worldwide votes
Additional Coverage from WIRED
Evaluations and Instructions
© 2024 Condé Nast. All rights reserved. Purchases made through our website may result in WIRED receiving a share of the sales, thanks to our affiliations with retail partners. Reproduction, distribution, transmission, caching, or any other form of usage of the content on this site is strictly prohibited without the express written consent of Condé Nast. Ad Choices
Choose a global website
Discover more from Automobilnews News - The first AI News Portal world wide
Subscribe to get the latest posts sent to your email.