An Australian startup called Baraja thinks it has figured out how to make the implementation of lidar in self-driving cars simpler, more robust and cheaper. How? Using prisms, actually, according to a piece by Wired.
Don't worry, we'll explain. lidar is a technology that uses lasers to map out an area visually. It's very good at providing an accurate picture of the world around it, but its downside comes when you need to move the laser to get a bigger picture. The most popular supplier of lidar scanners today, Velodyne, spins its lasers around in a housing almost 100 times per second but this has problems.
A system like the one Velodyne uses has a lot of moving parts, and when you mount something with a lot of moving parts on a vehicle, it's only a matter of time before the vibrations caused by driving shake it to bits. Baraja's system gets around this using prisms to bend the lidar's laser beam, dramatically reducing the number of parts in the system.
When you shoot a beam of light into a prism, the prism separates the beam into the rainbow color spectrum. Each color of light has a different frequency, Baraja uses this property to determine the angle at which the laser exits the prism, only the beam that Baraja us using isn't part of the visible spectrum, it's in the infrared range of light.
The Baraja sensor head is compact and robust. It connects to what Baraja calles the "engine" which drives the heads and processes all of the data from up to four sensors.
Baraja
Obviously, this is a super-simplified version of how the technology works, but Baraja's Spectrum-Scan could mean dramatically cheaper, simpler and more robust lidar sensors are now available to developers, bringing the cost of a self-driving car's sensor package down significantly, if Baraja can get enough developers to buy-in.
