Silicon Valley startup Aeva aims to give driverless cars better vision

In January, with funding from Lux Capital and other venture capital firms, two former members of Apple’s Special Projects Group, Soroush Salehian and Mina Rezk, started Aeva.

The company aims to improve the ability of self-driving cars to see their surroundings, according to New York Times report. Salehian and Rezk are reimagining Lidar—that is, Light Detection and Ranging—technology, which today’s self-driving cars use along with cameras, radar, GPS antennas, and other implements to create a picture of the world around them.

Aeva’s lidar, the company says, measures distances more accurately than other such systems. And, unlike other lidar systems, Aeva’s judges velocity. It is also smaller and less expensive than today’s lidar technology.

Aeva aims to have it on the market by 2018.

Traditional lidar devices emit pulses of light and measure their wavelength and return times to determine how far away a given object is. Then, computers use the data to construct three-dimensional models of the surrounding world.

But, today’s lidar systems can only detect objects that are relatively close, and cannot always differentiate between one object and another, the Times notes. As a result, they do not perform well in bad weather or when moving at high speeds.

Radar, which uses electromagnetic waves rather than light waves to map the world, can detect objects at greater distances, making it more suitable when traveling at high speeds, and cameras can “read” street signs and differentiate between, say, a pedestrian and a crosswalk.

So, cameras, radar, lidar and other devices work together to “drive” today’s autonomous vehicles. Driverless cars will likely continue to employ this combination for the foreseeable future, as multiple detection systems represent multiple layers of security.

Lidar devices, along with the rest of the ensemble, are expensive. It costs hundreds of thousands of dollars to outfit a self-driving car with the necessary hardware. The prohibitive cost of production prevents companies from marketing self-driving cars to average consumers. So, the first self-driving cars are not privately owned; rather, they have debuted in the fleets of companies like Lyft and Uber.

But, the Times cites a report by the Boston Consulting Group that projects that the self-driving car market will be worth $42 billion by 2025. For that to happen, companies must find ways to produce the vehicles more affordably.

The Times equates Aeva’s system to a cross between lidar—which is ideal for judging distances—and radar, which is best at detecting speed. Rather than emitting a series of light pulses, the device sends out a constant wave of light. This approach, Rzek told the Times, allows Aeva lidar to produce a better resolution, work better in inclement weather, and handle reflective surfaces better than standard systems do.

“I don’t even think of this as a new kind of lidar,” Tarin Ziyaee, co-founder and chief technology officer at the self-driving taxi start-up Voyage, who has seen the Aeva prototype, told the Times. “It’s a whole different animal.”

Researchers at the University of California, Berkeley, developed a similar continuous-wave lidar system back in 2014, the Times notes. Other companies that develop lira technology, such as Velodyne and Oryx Vision, are exploring similar options, according to said publication.

Lidar’s applications go well beyond driverless cars. Law enforcement uses the technology to create automated speed traps. Lidar may one day track a user’s movements for virtual-reality environments.

Today, video game systems like the Xbox Kinect do not use Lidar, because Lidar devices are too expensive, too bulky, and too power-consumptive for the purpose. But, continuous-wave Lidar systems are cheaper and lighter than pulse-based ones.

Behnam Behroozpour of U.C. Berkeley told phys.org in 2014 that he envisions that Lidar can be used for “a host of new applications that have not even been invented yet.” For instance, cell phones could use the technology to recognize a user and detect his hand motions from across the room, allowing him to control the device with simple hand gestures.

BMW’s first level-5 self-driving offering, which the company plans to release by 2021, will allow human riders to use hand gestures to order Amazon packages, make a dinner reservation, and perform a range of other actions.

Featured image via Wikimedia Commons

Uber and Google Battle for Footing in the Race to a Driverless Car

After announcing a partnership with Carnegie Mellon University in February, Uber has already started testing driverless car prototypes in Pittsburg. Carnegie Mellon has long been hailed as one of the top robotic engineering research establishments.

Taking that into consideration, Uber hired numerous award-winning researchers from the University in addition to other developers for this joint program. A Ford Focus was spotted on the streets of Pittsburgh this week sporting LiDAR sensors and an Uber Advanced Technologies Center badge.

LiDAR (Light Detection and Ranging) sensors have been used for mapping and navigation, but are now also being utilized for autonomous driving. The system can be mounted on the top of any car, and uses infrared lasers to analyze a car’s surroundings. It is used in combination with radar, GPS and computer vision technologies to ensure pinpoint location accuracy and instantaneous reactions to traffic incidents.

Having purchased the abandoned Restaurant Depot facility in Pittsburg’s Strip District, Uber has started to heavily invest in this project. Autonomous cars appear to be the future of the automotive industry. Uber has a steep hill to climb however, as Google has been testing autonomous cars since before the company’s launch date in 2010.

Google has already committed extensive resources toward developing the first autonomous car including prototypes in California and Nevada. The company has publicized this research broadly. The first and perhaps most notable PR move was a video of Steve Mahan, a legally blind older man, who was able to use a self-driving car to do errands around town.

Google has been testing self-driving prototypes for a couple of years now. With a fleet of 23 Lexus SUV’s, Google has logged more than 1 million miles of driverless car testing, and to this point has reported 11 minor accidents. According to Google, none of the 11 accidents were the responsibility of the system, rather the error of other drivers on the road.

Uber has a clear incentive to develop such technology further. Currently, the main cost of taking an Uber ride is the driver. This past Tuesday, Barclays’ Brian Johnson released a comprehensive study concerning the how much money the company could save if the need for a driver became obsolete.

“By removing the driver from the equation (the largest cost in a taxi ride), the average cost per mile to the consumer could be 44 cents for a private ride in a standard sedan and 8 cents for a shared ride in a two-seater.”

For an UberX customers currently pay $3 to $3.50 per mile. Such a drastic cost-cutting measure would revolutionize the transportation industry. With more than 160,000 currently employed by Uber, this technology will also be detrimental for the low wage work force, a population coming under increasing threat from a globalized labor force and hyper-efficient technological developments.

Johnson projected the autonomous car market to grow into a $42 billion industry in the next 10 years. While this may be on the high side depending on autonomous car R&D, this market is clearly the place to be for both technology companies and car manufacturers.

Driverless cars, once considered a futuristic fantasy, are becoming a question of when rather than if. Johnson predicts that 2025 will be a tipping point, further projecting that autonomous cars could compose a quarter of the entire automotive industry by 2035.

Image via PBS