May 7, 2019 by Guest Contributor
Originally posted on EVANNEX .
by Charles Morris
Tesla presented his latest and greatest vehicle autonomy features on his latest Investor Day Autonomy. The brain of the system is the Tesla Full Self-Driving (FSD) computer, which is now included in all Teslas products. The autopilot hardware package includes 8 cameras, 12 ultrasonic sensors, radar, GPS, an inertial measurement unit and sensors that measure the angle of the steering wheel and accelerator pedal.
|Tesla Camera and Waymo's lidar (Pictures via Tesla / Waymo )|
A feature that it does not have is lidar, a technology that uses pulsating laser light to measure the distance to a goals, and it is favored by most other autonomous driving companies (Uber, Waymo, Cruise). In its Autonomy Investor Day presentation, the eternal iconoclastic muscle rejected the lidar in shape reminiscent of those he used to describe hydrogen fuel cells. Lidar is "an idiot is erratic," he said. It is "expensive" and "unnecessary" and "everyone who trusts the lidar is doomed. It's like having a whole lot of expensive attachments. As one attachment is bad, well now you have a whole bunch of them. That's ridiculous As you TechCrunch candidates for lidar report their ability to look through adverse conditions like rain, mist or dust better than cameras. Disadvantages are that it is expensive and consumes a lot of power, and so Tesla's system is dependent on cameras. Andrei Karpathy, senior director of AI at Tesla, explained that visual recognition gives a better picture of the real world – he said lidar systems are having trouble distinguishing things like the difference between a plastic bag and a rubber tire. "That way, suffering is really a shortcut," said Karpathy. "It sidestep the fundamental problem, the important problem of visual recognition, which is necessary for autonomy. It gives a false sense of progress, and ultimately is a crutch." of vehicle information recorded by the thousands of Autopilot-equipped Tesla vehicles on the road. By using different AI techniques, Tesla learns his system to recognize and respond to the vast array of situations that can arise in nature.
"Everybody is training the network all the time," said Musk. "If Autopilot is on or off, the network is being trained. Every mile run for the car that is [equipped with Autopilot Hardware version 2 or above] trains the network."
Anthony Levandowski, former Google / Waymo engineer , running on lidar and saying "Elon is right." (YouTube: TechCrunch )
Some autonomous experts agree with Musk. As Gizmodo Cornell researchers report that cameras can rival lidar if properly mounted. Lidar systems are designed to provide a 3D image of a vehicle's environment and the road ahead, and therefore they are often mounted on top of a vehicle for the best viewpoint. In a paper presented at the upcoming conference on data vision and pattern recognition, the Cornell team explains that a pair of cheap cameras mounted behind a car's windshield can produce stereoscopic images that can be converted to 3D data almost as precisely as generated by lidar, to a fraction of the cost.
Mashable reports that British startup Wayve is a member of the camp that is not a member. The company says it doesn't need more data sources – just a GPS system, cameras and a powerful computer are all it takes to teach cars to drive as well as humans.
Editor's Note: ARK Invest's Autonomous Driving Expert has also told CleanTechnica on a recent podcast that they have come around to Elon Musk's point of view as well. Listen here:
Other players do not agree. A Chinese autonomous driving company called AutoX deployed a camera-based self-drive system in 2017, but has now added lidar sensors to their vehicles for redundancy and additional entry.
Musk is not quite the opposite lidar – SpaceX uses it in some applications *. For auto-autonomy, Musk believes that Tesla's sensor suite and trove of driving data will be quite adequate.
* Editor's Note: Musk noted during Autonomy Day that he led the development of a SpaceX lidar system for docking to the International Space Station.