A solid-state Lidar chip works by emitting laser light from an optical antenna. A tiny switch turns the antenna on and off. The light reflects off the sample and the same antenna captures it. For 3D imaging, the Lidar chip has an array of such antennae. The switches sequentially turn them on in the array.
An array of MEMS switches for high-resolution solid-state Lidar can reduce its cost significantly. This allows the solid-state Lidar to match other inexpensive chip-based radar and camera systems. This removes the major barrier to the adoption of Lidar for autonomous vehicles.
At present, autonomous highway driving and collision avoidance systems use inexpensive chip-based radar and cameras as their mainstream building blocks. At the same time, Lidar navigation systems, being mechanical devices and unwieldy, are also expensive, costing thousands of dollars.
However, researchers at the University of California, Berkeley, are working on a new type of high-resolution Lidar chip and this may be the game-changer. The new Lidar uses an FPSA or Focal Plane Switch Array. This is a matrix of micron-scale antennas made of semiconductors. Just like the sensors in digital cameras do, these antennas also gather light. While smartphone cameras have impressive resolutions of millions of pixels, FPSA on the new Lidar has a resolution of only 16,384 pixels. However, this is substantially larger than the 512 pixels that the current FPSA has.
Another advantage of the new FPSA is its design is scalable. Using the same CMOS or Complementary Metal-Oxide Semiconductor technology that produces computer processors, it is possible to reach megapixel sizes easily. According to the researchers, this can lead to a new generation of 3D sensors that are not only immensely powerful but also low-cost. Such powerful 3D sensors will be of great use in smartphones, robots, drones, and autonomous cars.
Surprisingly, the new Lidar system works the same way as mechanical Lidar systems do. Mechanical Lidars also use lasers for visualizing objects situated several hundreds of yards away, even when they are in the dark. They also generate high-resolution 3D maps for the artificial intelligence in a vehicle for distinguishing between obstacles like pedestrians, bicycles, and other vehicles. But for over a decade, researchers have tried to put these capabilities on a chip, without success, up until now.
The idea is to illuminate a large area. However, the larger the area illuminated, the weaker is the light intensity. This does not allow reaching a reasonable distance. Therefore, researchers had to make a trade-off for maintaining the light intensity. They had to reduce the area that their laser light was illuminating.
The new Lidar has an FPSA matrix consisting of tiny optical transmitters. Each transmitter has a MEMS switch that can rapidly turn on and off. This allows time for the waveguides to move from one position to another while allowing channeling the entire laser power through a single antenna at any time.
Routing light in communications networks commonly uses MEMS switches. The researchers have used this technology for the first time for Lidar. Compared to the thermo-optic switches that the mechanical Lidar uses, the MEMs switches have the advantage of being much smaller, consuming far less power, operating faster, and performing with significantly lower light losses.