Lidar-On-A-Chip Holds Promise For Low-Cost Autonomous Driving Systems — New Stanford Research

Lidar, shorthand for “light detection and ranging,” is conceptually the same as radar except it uses light instead of radio waves to “see” things not visible to the human eye. Lidar units do a better job of detecting living things like people and dogs than radar units, but radar is better at penetrating fog, smog, rain, and other atmospheric anomalies.

If you are designing an autonomous car, you might want to have both radar and lidar systems available, but there’s a problem or two. Lidar units today are big and bulky, which makes it hard to integrate them into the exterior of a vehicle. They also are frightfully expensive, costing upwards of $10,000 each — far too much for the ordinary motorist to spend.

Jelena Vuckovic. Image courtesy Stanford/Jelena Vuckovic

Jelena Vuckovic is a professor at the Stanford School of Engineering who has been involved with lidar systems for over a decade. Along with her research team, she has devised a new way to make a lidar system on a computer chip she says could be mass produced for a few hundred dollars. If her work turns out to be commercially viable, it could help usher in a new era of self-driving cars.

The breakthrough is a result of the fact that silicon is transparent to infrared light in much the same way that glass is transparent to visible light. Using a process called inverse design, the researchers have created an algorithm that sends a beam of infrared light outward in one direction and measures how long it takes to reflect back. That information helps reveal objects in the path of the beam. The next step in the process is teaching the “lidar-on-a-chip” to expand its field of coverage until it includes a full 360 degree circle without using the mechanical parts which greatly increase the cost of the finished product.

According to a Stanford blog post, Vuckovic estimates her lab is about three years away from building a prototype that would be ready for a road test. “We are on a trajectory to build a lidar-on-a-chip that is cheap enough to help create a mass market for autonomous cars,” she says. The results of the research to date were published March 23 in the journal Nature Photonics.

To date, Elon Musk has turned up his nose at using lidar, saying a suite of cameras and radar sensors can do a better job without the limitations of lidar. Perhaps he might change his tune if the cost of lidar units were to fall dramatically. For full self driving, it seems more information should always be better than less.