Search This Blog

Wednesday, November 29, 2017

Velodyne’s Latest LIDAR lets Driverless Cars Handle High-Speed Situations

Discerning a butterfly from a broken tire at 70 MPH.
As reported by The Verge: Self-driving cars from Alphabet’s Waymo are currently cruising the streets of suburban Arizona, navigating around with no human at the ready to take the wheel should something go wrong. It’s some of the most advanced testing we’ve seen so far, reaching what’s know as Level 4 autonomy. These cars can operate without any human input, but only under certain conditions and on certain roads.


Velodyne, one of the leading manufacturers of laser sensor for self driving cars, made an announcement this morning that it hopes will push things to the next level. The company released details of its latest product, the VLS-128. It’s the most powerful LIDAR the company has ever created, with twice the range and three times the resolution of its predecessor. “This product was designed and built for the level 5, fully autonomous, mobility as a service market,” says Anand Gopalan, the company’s CTO, meaning it can perform as well or better than a human under any circumstances.
On the bottom, the view from Velodyne's HDL-64.  On top, the more detailed view of the VLS-128.
The most important sensor in most self-driving cars these days is LIDAR, a laser scanner that can provide a 360-degree view of what’s happening around the vehicle. To illustrate the capabilities of the 128, Gopalan gives an example of a particularly challenging situation. “There is a small black object far out in front of you. Is it a piece of paper, a butterfly, or some tire debris? The autonomous vehicle needs to be able to see this object and make a decision about whether it should change lanes or break, and then take action. Traveling at 70 miles per hour, you have precious little time to do this.”


Gopalan says that the 128 can handle this sort of edge case. It’s 300 meter range and incredible detail are one part of the equation. But it also works for tricky scenarios like tire debris because it allows an autonomous driving system to take fewer steps between seeing the world and deciding what to do. “With lower-resolution LIDAR you would need to somehow fuse the data with cameras and do some processing to create something that can be understood by the computer,” he tells The Verge. “You now have such a high-resolution image, you can take the data and put it directly into an image classification algorithm. It reduces complexity and time.”
Just last month Nvidia claimed its computer vision systems are ready for Level 5 autonomy. Intel has made similar noises. And Waymo, in demonstrating its latest system to our reporter, touted its ability to see and identify debris on the road.

Of course, Velodyne doesn’t compete directly with either of these companies. Its LIDAR system is a complement to the chips sold by companies like Nvidia. And while some automakers are working on building their own LIDAR in house, most are turning to suppliers like Velodyne as they look to build driverless car services that would compete with the likes of Waymo and Uber.

Velodyne says that it managed to add more range and resolution to its latest unit, while simultaneously reducing the size, weight, and power consumption. For now, however, it’s staying mum on the price. In fact, says founder and CEO David Hall, price isn’t really the point. “We took a cost is no issue approach with this thing,” says Hall. “The mobility-as-a-service customer would just as soon have a higher end LIDAR. The costs aren’t that high, when compared to the value of not having a driver.”