CEO Elon Musk has said the laser-sensing device used in Googles driverless cars is not necessary. But after the fatal collision, will he reconsider his position?
The autopilot mode of the Tesla Model S, involved in a fatal collision made public last week, draws on a number of sensors including cameras, ultrasound and radar. But one piece of technology Tesla chose not to build into its cars is Lidar, a laser-sensing system used by Googles driverless cars.
Tesla CEO Elon Musk has, in the past, dismissed the need for Lidar, suggesting the technology doesnt make sense in the context of a car. For full autonomy youd really want to have a more comprehensive sensor suite and computer systems that are fail-proof. That said, I dont think you need Lidar. I think you can do this all with passive optical and then maybe one forward radar, he said during a press conference last October.
<
figcaption class=”caption” caption–img”> Googles self-driving car uses Lidar.
Will the recent accident where both the driver and the Model Ss sensors failed to spot a truck crossing the road ahead prompt Musk to reconsider that position?
Lidar, which stands for Light Detection and Ranging, consists of a cone or puck-shaped device that projects lasers which bounce off objects to create a high-resolution map of the environment in real time. In addition to helping driverless cars to see, Lidar is used to create fast, accurate 3D scans of landscapes, buildings, cultural heritage sites and foliage. It was also used to help create Radioheads House of Cards music video.
When positioned on top of a vehicle, it can scan up to 60 metres in all directions, generating precise 3D maps of the cars surroundings and ensuring the vehicle can avoid obstacles and collisions. Its expensive, but it provides visibility where other sensors can fail.
Lidar is the best of both worlds. It sits between cameras and radar and can detect both distance and objects, and can make out the shape of those objects, explained Richard Wallace, the director of the Transportation Systems Analysis group within the Center for Automotive Research.
Cameras and radar both used on the Tesla Model S are vulnerable to failings, he explains. The camera is an optical system like your eye. If its nighttime or theres a blinding snowstorm then cameras will struggle.
Radar, on the other hand, is good at detecting objects and how far away they are, but doesnt offer any information about the shape or size of the object. The radar in the Model S likely did detect the truck it collided into, but the system is designed to tune out objects that might look like overhead road signs to prevent it from false braking events.
They have to do that, otherwise imagine going down a highway and every time you come to an overpass it hits the brakes, said Wallace. Clearly the algorithm needs some refinement.
Although Wallace acknowledges that the Model S isnt designed to be fully autonomous, he suggests that Tesla may need to reconsider its stance on Lidar if its to achieve its self-driving ambition.
I know Elon Musk has said Lidar isnt necessary. Hes obviously a smart guy but at the end of the day I think its going to be proven that you do need Lidar in the mix, he said. It adds a level of resiliency and redundancy that makes the integration easier to solve.
The integration Wallace is referring to is the algorithms and intelligence that control the way the different sensors work together. All sensors have one shortcoming or the other. How do you create the brain that integrate them altogether and makes the right decisions?
Wallace believes that Lidar and vehicle to vehicle communication, where each car communicates their location to others nearby, will both play a key role in building safer self-driving fleets.
The Lidar units that Google uses in its self-driving cars cost up to $70,000 per unit, though there are now units that cost as little as $250. This could make it more accessible for the mass market.
Of course just having Lidar doesnt make a driverless car failsafe. Googles fleet has had its fair share of crashes and glitches, although no reported fatalities to date.
Tesla declined to comment, but pointed the Guardian towards Musks previous comments about Lidar not being necessary for driverless navigation. The company also flagged a list of factors outlined in the user manual of the Model S that can impair the performance of autopilot, including poor visibility, bright light (oncoming headlights or direct sunlight), damage or obstructions caused by mud, ice, snow and extremely hot or cold temperatures.
The list of limitations is followed by a warning that states: Never depend on these components to keep you safe. It is the drivers responsibility to stay alert, drive safely and be in control of the vehicle at all times.
The company also linked to a blogpost, titled Your Autopilot Has Arrived, which states: The driver is still responsible for, and ultimately in control of, the car. Whats more, you always have intuitive access to the information your car is using to inform its actions.