Intel-backed Perrone Robotics builds AVs without AI or cloud

Perrone Robotics recently finished a three-month trial of an AV shuttle on the streets of Crozet in Albermarle County, Virginia. (Perrone Robotics)

A small Intel-backed company in the autonomous vehicle (AV) space has automated 29 vehicles since 2003 without relying on artificial intelligence.

Perrone Robotics built a test track and headquarters in tiny Crozet, Virginia, nestled alongside the Blue Ridge Mountains. The company engineered a general-purpose robotics software platform called MAX that can be used to guide existing vehicles that are modified for autonomy to run along planned routes. They stop at signals and rely on instructions to avoid pedestrians and other obstacles when necessary. Numerous test runs, some on public roads, have shown the concept works.

MAX doesn’t need a big parallel processing chip from Nvidia and can run on a quad-core Intel chip, Perrone Chief Marketing Officer Dave Hofert said in an interview with FierceElectronics.

Free Daily Newsletter

Interesting read? Subscribe to FierceElectronics!

The electronics industry remains in flux as constant innovation fuels market trends. FierceElectronics subscribers rely on our suite of newsletters as their must-read source for the latest news, developments and predictions impacting their world. Sign up today to get electronics news and updates delivered to your inbox and read on the go.

“We’ve even run a demo on Raspberry Pi with two gigs of memory that cost $39 to show we didn’t need big processing power,” he said. The solution also doesn’t rely on cloud computing.

“We’re very different from competitors and not AI based,” he said. “All those AI companies are a bit stuck at the moment and that’s why autonomy in general has kind of slowed down. The AI solution is so complicated.”

Instead, Perrone’s MAX relies on intelligent routes and plenty of sensors for what a vehicle must do in different situations, such as stopping when it senses a stop sign ahead or parking or entering a roundabout. “All these essential modules are strung together in logical fashion,” Hofert explained.

On a practical level, it means that vehicles equipped with MAX software can be set up in hours instead of days to follow a demonstration route around a corporate headquarters, shopping mall or amusement park or to make trips to and from an airport. 

With a small team of fewer than 50 workers, the company boasts that it expects to deploy fully driverless vehicles in about a year for possible uses such as heavy AV trucks for hauling in a mining operation or shuttles for passengers.

Perrone has a handful of customers that are larger in size and conducting pilots. The company is currently entering a second investment round and hopes to use the proceeds to grow by ten times. Intel Capital provided the first round of funding three years ago.

Perrone’s vehicles are designed so far to operate completely independent of the cloud or street side infrastructure. Hofert acknowledged that “having some ability to talk to traffic lights or to get an indication of a car coming around an obstructed corner or when a pedestrian is about to cross a higher speed road will increase the reliability and effectiveness of our solution.”

With the use of well-known street routines like stopping at a crosswalk plugged into MAX, sensor technologies that are fused together are essential to Perrone’s success. The company relies on radar, LiDAR, camera and ultrasonic sensors installed in the vehicles it has deployed. “Having less expensive, more reliable and more accurate sensors will be key to broadly deploying AVs, both on the vehicle and in the environment,” Hofert said.

One of Perrone’s recent trials of its technology finished in October in Albemarle County, Virginia. Providing free public demonstration rides for three months, Perrone boasted the AVNU (Autonomous Vehicle, Neighborhood Use) shuttle drove 530 miles and carried 750 passengers without any safety problems. (A safety driver was on board, as required.) The Polaris GEM e6 shuttle used public roads, navigating vehicle, bike and pedestrian traffic through intersections and roundabouts.

Richard Windsor, an analyst at Radio Free Mobile, said he has studied many companies working in the AV development space and understands the complexities of working with AI, which is only as effective as the historical data it relies upon.

What Perrone does is sell the tools to enable an AV solution. Using MAX on a fixed route where data is stable will be effective, but Windsor said in an interview that he’s not convinced Perrone is close to taking an AV out on the open road.

Perrone’s processing power

Perrone’s engineers have often used a Nuvo computer from Neousys that runs on a quad-core Intel Core i7 in its vehicles because of its ruggedness and reliability. “As we move forward, we’ll need a rugged but strong processor board that is more compactly built, and we’ll need two for failover,” Hofert said. A third processor would handle side tasks not needed for direct control.

Perrone has also worked with the Xilinx Zync Ultrascale MP SoC processor, which is based on the Arm+ embedded GPU co-processor architecture. The company expects to use that processor on a future field trial.

“What ultimately matters is reliability and the ability to take extremes of temperature that you see in vehicles operating in the open,” Hofert said. “Plus, a compact size to fit into the various nooks and crannies you find on any vehicle.”

RELATED: Western Digital sees mega storage needs for AVs


Suggested Articles

In June, Su stood out for heartfelt commentary on social injustice after the killing of George Floyd. She challenges industry to do more.

New chip will be incorporated in coffee mug-sized device to cost $199 next year

In a nutshell, Edge ML is a technique by which smart devices process data locally using machine and deep learning algorithms.