Inside Aeva's Aeries system is a small chip that incorporates all the core optical components of lidar. (Image courtesy of Aeva).

Startup Says It Can Slash Price of Autonomous Car Sensors

Dec. 19, 2019
Aeva's lidar sensors are compact and could cost less than current models, which resemble coffee cans rotating on an autonomous car's roof, blasting beams of light out into the world and using the reflected light to tell the distance and depth of objects.

The autonomous cars being tested today track their surroundings using a broad range of sensors, ranging from radar to cameras to lidar, each one covering each other’s blind spots. The lidar sensor is one of the key components, acting as the omniscient eyes of the car and comparing what it sees with the other sensors. But the soaring cost of lidar is considered one of the biggest obstacles to the mass production of autonomous cars.

Aeva, the Silicon Valley startup founded by former Apple engineering leaders Soroush Salehian and Mina Rezk, is looking to slash the high price of current models. Aeva said it crammed all the core optical components of a lidar sensor on the same slab of silicon, which it hopes to sell for under a thousand dollars. The system can not only deduce the distance to objects, including cars and people, but also capture speed—not unlike radar.

Today, autonomous cars use these sensors to create three-dimensional dioramas of the surrounding area, alerting drivers to danger on the road ahead. Aeva's sensors are more compact and could cost less than current models, which resemble coffee cans rotating on the autonomous car's roof, blasting beams of light out into the world and then measuring how long it takes for them to bounce back to determine the distance and depth of object.

The spinning sensors introduced by global leader Velodyne Lidar can create 360-degree   coverage of the car's surroundings. The problem is that these complex contraptions can cost several thousands to tens of thousands of dollars each. Aeva and other startups are trying to solve the problem by selling so-called solid-state devices capable of scanning out over long distances and locating objects without whirling cylinders or consuming lots of power.

"One of the biggest roadblocks to bringing autonomous vehicles to the mainstream has been the lack of a high-performance and low-cost lidar that can scale to millions of units per year," Salehian, one of the company's founders, said in a statement. He said that its solution can be manufactured at "silicon scale." Aeva agreed in April to sell sensors to the Audi unit building autonomous cars for Volkswagen, the world's largest car manufacturer. 

Alex Hitzinger, who leads the autonomous driving division at Volkswagen, believes it is “the best lidar solution on the market.” He said that Aeva's sensors solve “a fundamental bottleneck for perception in taking autonomous driving to mass scale.” The Silicon Valley company, which has more than $45 million in venture funding including from Porsche SE, Volkswagen's largest shareholder, plans to start selling its latest lidar sensor in early 2020. 

Volkswagen plans to roll the sensor out to new cars in 2022 or 2023.

Aeva believes it can sell sensors that resolve objects up to 300 meters ahead and cost less than $500, in contrast to current models that cost many thousands of dollars. The sensors are also able to withstand interference from sunlight and lasers used by other lidar sensors, which can scramble the sensor's vision. Aeva's systems can also be used to spot objects that can reflect only small amounts of light, ranging from lane markings to street signs.

While rival lidar sensors release flashes of light into the surrounding world, Aeva's system shoots continuous beams that change frequency over time. The sensors can figure out the coordinates of objects, ranging from cars to construction equipment on the side of the road, by measuring the change in frequency after the beam returns. The process is called frequency modulated continuous wave, or FMCW, which can also be used to judge speed.

By using the difference between the frequency of the departing and returning light, the sensor can not only measure distance more accurately but also capture the speed of an object and predict where it's headed, according to Aeva. That results in higher resolution perception, which is indispensable for telling a child darting out into the road from a street sign. Radar sensors also use the changing frequency of radio waves to calculate the speed.

Aeva’s Aeries sensor system is capable of figuring out the distance and speed of objects within its 120-degree field of vision, which means that at least three sensors are needed to create 360-degree coverage. But because it is half the size of its predecessor, the unit can be slapped above headlights, mirrors and bumpers. Aeva said the sensor could be used to enable both autonomous cars and other functions such as lane-changing assist.

“Not all FMCW lidars are created equally,” Rezk, the company's other founder, said in a statement.

About the Author

James Morra | Senior Editor

James Morra is a senior editor for Electronic Design, covering the semiconductor industry and new technology trends, with a focus on power electronics and power management. He also reports on the business behind electrical engineering, including the electronics supply chain. He joined Electronic Design in 2015 and is based in Chicago, Illinois.

Sponsored Recommendations

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!