Published on May 23rd, 2018 | by Steve Hanley0
Mobileye Does More With Less, Receives Order For 8 Million Autonomous Driving Systems
May 23rd, 2018 by Steve Hanley
Mobileye was just another startup you never heard of until the day Joshua Brown was killed on a Florida highway in a Tesla while using Autopilot. That led to a rather nasty spat between Tesla and Mobileye, after which both companies went their own separate ways. Since then, Intel paid more than $15 billion to acquire the Israeli company. Today, many of the autonomous driving systems offered by world manufacturers — Nissan, Audi, Cadillac, BMW, Honda, and Nio, among others — are based on MobilEye technology.
Order For 8 Million Systems
CEO Amnon Shashua tells Reuters there are currently 27 million cars on the road from 25 different automakers that use some driver assistance technology — which could be as simple as a forward collision warning or a blind spot monitoring system — and more than 70% of them are sourced from Mobileye. “By the end of 2019, we expect over 100,000 Level 3 cars with Mobileye installed,” he says.
The company has just received an order for 8 million autonomous driving systems from an unnamed European manufacturer. The contract calls for the first deliveries to begin in 2021 when the company’s most advanced EyeQ5 chip will be available. MobilEye’s EyeQ4 chip is expected to begin shipping in the next few weeks. Not every car fitted with a Mobileye chip will offer the full range of assistance the chip is capable of.
Robo-Taxis In 2021?
Shashua says that self-driving taxis should start becoming available in 2021 or soon thereafter. “When designing our system. we are looking at all what can be used today, in a year, in two years and then the robo-taxi,” Shashua said. It’s possible some luxury cars and a few less expensive cars will make such systems available at about the same time. The systems will add about $12,000 to the price of those vehicles, he says.
When self-driving cars start sharing the road with conventional cars, there will be safety issues, Shashua admits, and self-driving systems will be under pressure to outperform human drivers by a significant margin. Today, there are 40,000 vehicle fatalities a year in the US. He thinks the public will not tolerate even a 50% reduction in that number. His thinks something like 40 deaths will be acceptable, a decrease of over 99%.
12 Cameras. No Waiting.
There is no manual on the best way to build a self-driving car. Everyone is trying to figure it out as they go. Some like Uber have bulbous appendages on their roof that look like the robot from Lost In Space. Perhaps when they spot another car on the road, they send a coded message to a display inside the car that says “Danger, Will Robinson!” Elon Musk was so annoyed by the Joshua Brown fatality, he tore up the entire system built in cooperation with Mobileye that relied primarily on cameras and reconfigured it to make radar the primary input.
Musk says every Tesla built today will be capable of full autonomy once the software catches up with the hardware. Maybe. Musk refuses to use Lidar, a sensor similar to radar that relies on light rather than radio waves. Musk says its because Lidar does a poor job of peering through fog, smoke, snow, and rain. But Lidar is also ungodly expensive right now, so perhaps cost has something to do with his decision. He may also object to having bulky Lidar sensors interrupting the clean, flowing lines of his cars.
Mobileye has built a self driving system that utilizes nothing but cameras — no radar, no Lidar, although it does get input from a GPS unit. 3 cameras are forward looking with 3 different focal lengths. 1 looks to the rear. There is a forward and rear facing camera on each side of the car. 4 more are installed around the perimeter of the car. Each camera costs no more than $20. As CNET Road Show says, “You could outfit a whole fleet of cars for what Velodyne will charge you for a single Lidar scanner.” Here’s a video of the system in operation on the streets of Jerusalem.
Driving in Jerusalem is not for the faint of heart. Often, a driver needs to force the car into a tiny opening in traffic — or create an opening where none exists. The Mobileye system has been programmed to have just the right amount of assertiveness to fit in with its surroundings. The level of aggressiveness is defined by what the company calls Responsibility Sensitive Safety, or RSS. In the real world, while operating among human drivers, a car must have some amount of aggression to survive. RSS defines that boundary between too conservative and too risky, according to CNET. Here’s a video of the car negotiating traffic in Jerusalem and applying a healthy dollop of aggression when needed.
Even though its camera-based system works, it is not sufficient for full Level 4 or Level 5 driving. So Mobileye is working on a separate system that uses radar and Lidar but no cameras. Once that system if perfected, the two technologies will operate in parallel, providing a degree of redundancy that is not possible with a system that uses input from all three sources all the time.
“Once you have a comprehensive solution with only cameras, then you can start pin-pointing where you need additional sensing,” Shashua says. “You are now in a better position to fine-tune exactly where you need a three-sensor modality, and where you need only two… If you fuse from the beginning, for every angle you’ll need all three.” In theory, a car will be able to navigate using either one of the systems. If one fails, the other can take over to get the driver safely to the destination.
Speaking of failure, the sensors Mobileye is developing have a “mean time between failure” or MTBF, of a billion hours according to Engadget, which means the sensors can be expected to operate properly for that long of a period of time before needing replacement,
Mobileye plans to use Valeo Scala Lidar sensors that can be incorporated into the bodywork of a car where they will be largely invisible to people looking at the car from outside. They have only a limited field of vision, so a number of them must be used on every car, but at cost of a few hundred dollars each, all the sensors needed for one car will be less costly than one conventional Lidar unit from a traditional supplier today.
A Swipe At Tesla?
Redundancy ranks high in the hierarchy of things any autonomous driving system must have, according to Shashua. He tells CNET Road Show, “The way that you guard yourself against sensing mistakes is not only through algorithms that are well tested, but also through redundancy. So now, the level of risk you are willing to take is comparable to the level of redundancy that you are willing to invest in.”
For him, Lidar is an essential part of the package, but Tesla says it can have full redundancy without it. “So what is Tesla stating? They are willing to take more risks in sensing mistakes because they have less redundancy, because they want to have a lower-cost technology? Another car manufacturer can come in and say, ‘I want to have almost zero risk in having a sensing mistake and I’m wiling to pay more for that, or my customers are willing to pay more for that, so therefore I’m putting not only radars and cameras, but also lidars.’ I wish them the best of luck moving forward,” he adds.
One Horse In A Crowded Field
As far along as Mobileye is with its autonomous technology, it is in a horse race with dozens of other companies, many of them in China. In the US alone, Nvidia, Uber, and Waymo are champing at the bit to get self driving cars on the road. And then there is Tesla, which insists it will have a car go from LA to NYC without a human hand on the wheel before the end of the year.
The payoff is money — lots of it. Ride-hailing and mobility services are projected to be worth $500 billion a year in China alone within 10 years. Between now and then, many systems will be tried — systems that draw a fine line between safety, appearance, and cost. Lidar technology is making rapid strides in terms of cost and size. Waymo says it has slashed the cost of its Lidar sensors by 90%. Sometime during the coming years, there will be convergence between the various manufacturers on what hardware is required and what can be omitted. Industry standards will begin to emerge.
In the meantime, it’s the wild, wild west in the field of autonomy. Mobileye could emerge as an industry leader, or some company you never heard of could zoom into the lead as others fall by the wayside. The only thing that could slow the pace of progress is if self-driving cars start getting into more and more fatal accidents. Ammon Shashua is right. Public acceptance of fatalities where autonomous systems are in use will be near zero. Companies like Uber, which decided to reduce the number of Lidar sensors installed on its self-driving test vehicles to save money, could set the whole idea of autonomous driving back decades with its foolish disregard for safety.
Follow CleanTechnica on Google News.
It will make you happy & help you live in peace for the rest of your life.