The Large Hadron Collider technique could help self-driving cars "see" the road

The Large Hadron Collider technique could help self-driving cars “see” the road

Part of the Large Hadron Collider (LHC) complex is seen underground during an open day at the CERN particle physics research facility on September 14, 2019 in Meyrin, Switzerland.
Photo: Ronald Patrick (Getty Images)

a three-year of study between from Volvo independent software arm and The Large Hadron Collider in Europe shows promise in helping self-driving cars understand the road and make shared decisions.

The Large Hadron Collider is undeniably epic. Iis the world’s largest and most powerful particle accelerator capable of crushing particles at close range speed of light using superconducting magnets. After the collision, physicists take a look at the extremely short-lived wreckage to discover the building blocks of our universe. The LHC is run by CERN, or European Nuclear Research Council and accelerate particles since 2008.

With extensive experience studying accidents, CERN is uniquely positioned to help car companies develop machine learning programs capable of split-the second analysis of complex situations. The subatomic particles studied by CERN researchers are infinitesimally small and only exist for a microsecond. To understand what these particles are, CERN has developed algorithms that use Deep Learning to identify these fragments of matter.

The researchers have spent the past three years working with Volvo’s self-driving software development subsidiary, Zenseact, focusing on computer vision, which includes all the programming that allows computers to “see” and react to the real world. The goal of the research was to use the LHC’s deep learning algorithms to improve decision-making in autonomous cars.

“The results of our research show that there is still room for improvement when it comes to running deep learning algorithms faster and more energy-efficiently on resource-constrained hardware devices,” said Christoffer Petersson, director of research at Zenseact. in a press. release. “More simply, machine learning techniques could help make decisions faster in self-driving cars.”

Zenseact wanted to know if its chosen chips, known as Field Programmable Gate Arrays, could handle complex algorithms. From press release:

The main result of the FPGA experiment was a practical demonstration that automotive computer vision tasks could be performed with high accuracy and short latency, even on a processing unit with limited computing resources.

“The project clearly opens up future research directions. The developed workflows could be applied to many industries, for example, automotive.” Christopher explained.

Many of the challenges facing future scientific experiments and the technological challenges of the automotive industry require the processing of large amounts of data in real time, often by high-end computing devices with strict latency and power consumption constraints.

The joint research team of Zenseact and CERN carried out this project in an open-source software environment. The collaboration reveals that the world’s largest physics experiment could clearly help self-driving. The results show great promise for future increases in speed and accuracy in image recognition for self-driving vehicles, helping to improve cars’ ability to avoid accidents. For CERN, it was also a fruitful collaboration.

The research didn’t just benefit from Zenseact. Researchers at the LHC were able to refine their algorithms during the study as well.

“Our work together has elucidated compression techniques in FPGAs that could also have a significant effect on increasing processing efficiency in the LHC data centers,” Maurizio Pierini, a physicist at CERN. “With machine learning platforms setting the stage for next-generation solutions, the future development of this research area could be a major contribution to many other fields beyond high-energy physics.”

Volvo claimed in 2022 that it was on the verge of bringing a self-driving car to market with or beyond Level 3 autonomy this year. Zenseact, together with LiDar development company Luminar Technologies, developed the autonomous driving system called Ride Pilot. The subscription service should be available to drivers in California sometime this year, but Volvo hasn’t mentioned Ride Pilot much since then. The 2022 Consumer Electronics Show. We’ve reached out to Volvo about how this research affects Ride Pilot and will update this story when we learn more.

Leave a Comment

Your email address will not be published. Required fields are marked *