Why is LiDAR required for DRIVERLESS CARS?
These are interesting times, Driverless cars or Autonomous Vehicles are slowly entering the massive global automotive industry and it is important for us to know how different technologies come together to make driverless cars a reality. Here, we take up a series of technologies involved in what comprises a driverless car and the topic of discussion in this post is the LiDAR. This will answer your questions on how an Autonomous car sees on road and what is the technology which enables it to do so.
Just like the Radar (Radio Detection and ranging) uses radio waves to determine the range, angle and velocity of objects and Sonar (Sound Navigation and ranging ), the LiDAR for driverless cars function based on the same concepts.
LiDAR (Light Detection and Ranging) is basically a cone or puck shaped device mounted on roof, projecting laser lights which eventually bounce off objects it encounters and generates a precise, high resolution 3D map of the surrounding environment within which the vehicle can be driven. The bouncing back of lights, generates a point of cloud at a rate of a million points per second and are then interpreted, generally in the form of polygons. Through this methodology, we would be able to know about the obstacles in the path of a driverless car like lanes, traffic lights etc,. LiDAR is a very important part of driverless cars and significant development has been seen in this area (and more is expected ). Modern LiDAR’s have the ability to even detect and differentiate between human and bikes, their speed and direction.
Lexus Modified by Google Source (Google)
LiDAR mounted on roof top ( Source: Voyage . auto)
The LiDAR usually is positioned on the top of the vehicle for a complete 360 degree view from the pack of laser units and can be easily identified in a driverless car based on its position in the vehicle. An Average range of a LiDAR can be between 1000 ft to 2000 ft.
A LiDAR generated map Source(Voyage.auto)
With the clarity in the objective of using LiDAR, certain companies like Tesla have still not associated their cars with it and completely rely on the usage of cameras. BTW there are talks that Cameras can be fooled by bright sunlight and can become difficult to operate in storms etc. Thus, LiDAR shows a clear advantage over cameras but still not are able to completely make their way due to its economic feasibility. As of date an huge effort is going on to reduce the costs. Recently a Canadian company got funded with $101 Million to develop low cost solid state LiDAR’s that can be commercially available as low as $200 per unit ( Check more info on this HERE )
With the advantages of LiDAR’s discussed above, it’s time to know about its weakness as well. The important one is that it cannot see behind solid /opaque objects. For eg. if a bird sits next to LiDAR, a wide range of vision would be blocked. Interference is another issue which can be engineered to perfection.
Coming to the cost, as discussed earlier it is quite expensive. The Google Waymo Driverless car project has brought down the cost by almost 50 % and still a lot of on-going development is happening. This sort of signals a promising future for lessening the cost in market soon. Also, the Solid State LiDAR, which is currently the topic of discussion in the industry can reduce costs by 95% and startups like Leddartech are working to make it economically feasible.