I have written before that Mobileye, a company owned by Intel since 2017, is among the most awesome players in the self-driving industry. The Israeli company is a leading provider of chips, cameras, and powerful software for modern driver assistance programs – a few years ago, Mobileye demanded a 70 percent market share. The company hopes to enjoy the same high position in the growing market for full-fledged self-driving programs.
On Monday, Mobileye announced that its automotive technology stack would be called Mobileye Drive. Mobileye said the technology would be “a system of automated transactions that are suitable for commercial deployment at a standard level.” The Mobileye Drive system will have 13 cameras, three long-distance lidars, six short-wave lidars, and six radars. It will be powered by Mobileye’s EyeQ 5 processors.
Mobileye said the technology would be ready for commercial use by 2023. That would be a great thing if it were true, but I can’t doubt it. Over the past five years, many large self-driving companies have announced their launch dates and failed to meet them.
If Mobileye can prepare the technology for sale in 2023, he has the potential to become a major player in the emerging driving economy. Even if Mobileye loses its 2023 target, the company can still be a key player when it finally enters the market.
Mobileye already has strong relationships with major automakers, and Mobileye plans to market Mobileye Drive and nonautomakers, too. On Monday, Mobileye announced the sale of the Mobile Mobile Drive in the start of the launch of Udelv. Delv says it plans to use at least 35,000 delivery robots between 2023 and 2028 provided by Mobileye’s technology. Mobileye hopes this will be the first of many sales for the Mobileye Drive platform.
The power is different
Mobileye began testing in Munich in 2020.
Growth / Mobileye began testing in Munich in 2020.
Over the past five years, many companies have announced plans to roll out commercial automotive technology in 2018, 2019, 2020, or 2021. Few (if any) meet their deadlines.
But Mobileye says it’s different. The company says it has never made unfulfilled promises in the past and is on track to bring full driving technology to customers before the end of 2023.
Mobileye has an in-depth benchmark of engineering talent from its years of building Hardware and Driver Assistance Software. Mobileye is also in the process of launching one of the industry’s pilot programs.
Mobileye tested its expertise in Israel and Germany last year, and the company is working to expand testing in Detroit, Tokyo, Paris, Shanghai, and New York City this year. (Not by accident, some of these cities are also the headquarters of major car manufacturers.) Mobileye says testing in various cities is important to ensure that its software does not exceed the quirks of a particular city.
That has to do with one of the benefits of Mobileye’s strategy: access to gobs of map data. Mobileye has signed deals with several of his colleagues who work to collect information on cameras in customers’ cars as they roam the cities around the world. Mobileye says he has primarily made the process of converting this image data into more accurate, three-dimensional maps covering most of the world’s roads.
These maps will not only contain information about road geometry – they will also provide important information about driving patterns that will help Mobileye’s vehicles adapt to the existing traffic.
A different approach
Many self-driving companies use a process called sensor fusion: they take inputs from cameras, lidar, radar, and other sensors and combine them into a single integrated global model. This model is then transferred to a programming module that outlines what the vehicle should do.
In contrast, Mobileye made two fully integrated systems – one operated by cameras and the other a combination of lidar and radar. Once each system has achieved high performance separately, Mobileye will integrate them into one system. Mobileye believes that this non-compliance will give the company more confidence in its services.
There is a lot you can like about this strategy, but it is clear that it will work. Mobileye continues to rely on a head-to-head maths debate to help secure the safety of its technology. Mobileye says – if it can show that its camera-based system can travel 10,000 hours between a crash and its lidar-based system can run 10,000 hours between crashes – its integrated system will be able to travel 100 million hours (10,000 times 10,000) without crashing .
This argument is entirely assumed that the methods of failure of both systems are statistically independent. That doesn’t seem like logical thinking: situations that confuse one system may confuse another. So you can’t just repeat two chances together.
Finally, it is hard to see how anyone – even those within Mobileye – can hope that the company’s technology will be ready by 2023. With the company’s own adoption, Mobileye’s camera-based system has grown significantly beyond the lidar-based system. The company is still gathering information that it will need to convince regulators and the public that its system is secure. Mobileye is likely to be on track to bring full-time driving technology by 2023. But we will not know for sure until it happens.
Mobye’s partnership with car manufacturers will make it profitable by selling its technology to car manufacturers. But experts expect that most cars will not belong to consumers. Instead, they will be on ships used for taxi demand and delivery services.
On Monday, Mobileye announced an agreement to supply the upcoming self-driving technology