LiDAR, the Technology to Make Self-driving Cars a Reality

Cadillac SuperCruise 6 photos
Photo: GM
Cadillac CT6Cadillac CT6Tesla Model 3Tesla Model XVolvo with LiDAR

During the 2005 DARPA challenge, engineers from the Stanford University came closest in achieving the decades-old dream of a self-driving car, as envisioned in staples of the sci-fi genre like TimeCop. To make out the world around it, their vehicle used something called LiDAR, a technology never before implemented on cars. 

Arguably the first commercial large-scale application of the technology for the automotive industry came in 2018, when GM used LiDAR sensors mounted on trucks for accurately 3D mapping the entirety of the U.S. highway system. The data was to be fed into their Cadillac CT6, and later the CT5 and CT4, in order to allow these to navigate our nation’s highways all by themselves, with pinpoint accuracy.

Cadillac CT6
Photo: General Motors

You might have gathered from above that the main piece of technology responsible for the newer Cadillac’s hands-free driving isn’t actually part of the car itself. Its price is still too prohibitive for production cars, even if we are talking Cadillac. The LiDAR most Caddies do carry today is something different, although it works through the same basic principle.

LiDAR stands for (Light Detection and Ranging) and is a detection device that uses laser beams to calculate the distance to an object. When the light collides with the object, part of it is reflected back toward the source and picked up by a sensor. Range is determined by measuring the time interval between the emission and the return.

Considering that light travels at almost 200,000 miles per second, the level of precision required to apply this principle is astounding. Harder still to believe, LiDARs have been around since the 60s (famously used to map the surface of the moon), and the best commercially available models today can achieve a resolution of under an inch.

Cadillac CT6
Photo: General Motors
The ubiquitous collision sensor is nothing more than a range finder, which sends one or two beams in front of the vehicle to determine the presence of an object, its distance and relative speed. Most are not sensitive enough to differentiate between shapes, while smaller obstacles such as pets might as well not be there.

A modern mapping LiDAR will send millions of laser beams per second in all directions, and then feed the sensor information to a computer to create a live map of the environment.  

With enough computing power behind it, a 3D LiDAR can be used to track moving objects the size of a pet, predict their heading, differentiate between potential obstacles in the environment, turn alongside a pronounced curb, etc.

Volvo with LiDAR
Photo: Volvo
The biggest initial hurdle to implementing LiDAR in cars was low scanning speed. Early devices just could not build a map fast enough for the vehicle to move at anything else than a snail’s pace.

Come David Hall, and his invention of a “3D laser-based real-time system”. Unlike previous designs, which employed only two beams to slowly “brush” over surroundings, Hall’s solutions allowed for multiple directions to be scanned simultaneously. Emitters sent beams around a 360-degree arch, at a frequency high enough to paint a functional picture in nanoseconds.    

Hall’s invention was patented in 2005, and his company, Velodyne became the first commercial manufacturer of car borne LiDar systems, and they remain at the top of the industry hierarchy – such as it is. 

Tesla Model X
Photo: Tesla Cars
For the past 15 years, the company has been actively working to improv LiDAR technology, especially the size and cost issues that currently make the system unattractive for the car industry.

Some innate deficiencies might never be solved, however, like the LiDAR’s inability to “see” anything other than shape – an obvious impediment when driving.

A future self-driving car will probably require camera assistance for distinguishing road markings and signs, as chances are the system will never be self-sufficient. Conversely, it is becoming increasingly clear that the same may apply to competing technologies.

If you liked the article, please follow us:  Google News icon Google News Youtube Instagram X (Twitter)

Would you like AUTOEVOLUTION to send you notifications?

You will only receive our top stories