Stanford, Purdue Project May Improve AV Laser Steering Tech

The laser scanning technology keeping autonomous vehicles on the road could soon get a major upgrade, thanks to the work of researchers at Purdue University and Stanford University.

The research team believes they have invented a laser light sensing technology that more accurately guides a self-driving vehicle — an innovation made possible by scanning at higher frame rates and improving image resolution.

Such ultrafast laser beam steering for autonomous vehicles could vastly improve responsiveness, a key component to automakers seeking Level 5 autonomy — the level where an AV can operate on any road and in any conditions a human driver could navigate.

Self-driving vehicles currently rely on Lidar — similar to radar — but it’s dependent solely on light detection and ranging, an expensive and conspicuous solution (Lidar is housed in the “hats” that current AVs wear).

The new chip-compatible technology, jointly owned by Purdue and Stanford, doesn’t require additional sources of energy to run, while the use of photonic metasurfaces provides a more compact — read streamlined — and power efficient solution.

The researchers claim this innovation is “orders of magnitude faster” than conventional laser beam steering devices, which employ phased antenna-array technology.

In current phased-array optical technology, each antenna needs to be controlled in what it radiates individually, but with the newly developed system from Purdue and Stanford researchers, each of the structures emit slightly different frequencies.

This removes the need to address each individual antenna continuously, which in turn reduces overall power consumption.

“This technology is far less complex and uses less power than existing technologies,” Amr Shaltout, a post-doctoral research fellow in materials science and engineering at Stanford, who also conceived the idea of merging the metasurfaces with ultrafast optics, wrote in a statementannouncing the project’s results.

This allows the laser to scan a large angle of view in nanoseconds or picoseconds (compared with the microseconds current technology takes), resulting in more precise and responsive steering.

“The idea proposed by Amr is so powerful that we were honestly surprised that nobody did it before because it is so simple, so efficient, much easier than what people used so far and works much faster,” Vladimir Shalaev, the Bob and Anne Burnett Distinguished Professor of Electrical and Computer Engineering at Purdue, wrote in a statement. “This as a wonderful example of fruitful collaboration between Purdue and Stanford.”

The team is now looking for investors, partners or possibly licensing agreements as they work to scale up the technology and move it from the lab into the real world.

Possible initial real-world development applications include scanning devices at stores or airports, before development aimed at autonomous cars and OEMs building advanced drive assistance systems (ADAS) features.

“Sometimes working outside our field help us to see, to find solutions to problems in different fields and just link them together,” Shalaev wrote.

With the recent fatal accident involving a self-driving Uber vehicle in Arizona, all industry players — from Toyota to Tesla — are looking to assuage public fears that robot cars are dangerous not to be trusted.

“This seems to be a disruptive solution which could make a big difference in this huge, emerging industry,” Shalaev added.

— Nathan Eddy is a filmmaker and freelance journalist based in Berlin. Follow him on Twitter @dropdeaded209_LR.

Leave a comment

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