Volvo says it has managed to drive down pricing of LiDAR technology to levels that will enable use in self-driving cars as soon as 2022. This could represent a revolutionary advance in autonomous vehicle technology as LiDAR, often seen in prototypes made by Uber, Waymo, and other companies, has so far been prohibitively priced – at levels that makes commercial usage impossible.
However, it now seems Volvo’s 2018 strategic investment in Luminar, which makes sensors for autonomous vehicles, has paid off. According to Volvo, Luminar’s ‘perception’ technology will work alongside cameras and radars to allow cars based on the upcoming SPA2 platform to be ready for autonomous operation by 2022.
Volvo will initially offer a LiDAR-based Highway Pilot feature that it claims will enable fully autonomous highway driving. This will allow drivers to, in Volvo CTO Henrik Green’s words, “take your eyes off the road and your hands off the wheel” on highways once the car determines it is safe to do so. Green also revealed that Volvo would use over-the-air updates to add new capabilities to vehicles and to expand areas where completely autonomous driving is possible. “A safe introduction of autonomy is a gradual introduction,” he added.
While LiDAR systems are already found in a few cars (Audi uses a system made by Valeo in a few premium models), these are reportedly not powerful enough to enable usage at highway speeds. The biggest issues here have been cost and size, with the more powerful roof-mounted units you’d see on self-driving prototypes running into thousands of dollars. And that’s after a nearly 90% fall in pricing (from $75,000 in 2009), as Waymo CEO John Krafcik explained at the 2017 Detroit Auto Show. Luminar, meanwhile, seems to have pushed costs down further, and has previously stated it could manage to provide LiDAR systems for less than $1000.
Why so much fuss about LiDAR?
Most carmakers and self-driving startups seems to be relying heavily on LiDAR, which uses pulses of light to sense objects: Luminar says LiDAR can provide more reliable ‘perception’ as compared to just radar and cameras, while Waymo cites the accuracy of its LiDAR in improving safety – it’s capable of not just identifying pedestrians but also detecting what direction they’re facing in.
LiDAR also has the advantage over camera-based systems in low or tricky light conditions, and is less prone to being fooled by shadows. At the same time, critics of LiDAR point to its cost, its inability to distinguish colors and ‘read’ road signs, as well as the difficulties it can face in poor weather conditions, to insist a combination of sensors is key. Tesla, of course, is famous for rejecting LiDAR, with founder Elon Musk describing the focus on LiDAR as a ‘fool’s errand’. Instead, Tesla uses a camera-based computer vision system to achieve ‘pseudo LiDAR’, with advanced real-time processing of camera imagery used to predict distances of objects.