Delphi and Mobileye collaborate to build CSLP autonomous driving platform

Delphi and Mobileye, as the industry's leading parts suppliers, may not be as well-known as Google and Tesla in the field of autonomous driving. Delphi, originally a subsidiary of General Motors that provided parts, has been emerging in the field of autonomous driving technology for three years and is now headquartered in the UK. Mobileye is one of Israel's largest technology companies, specializing in vehicle collision avoidance technology and visual processing. The two companies believe that the secret to their ability to beat other competitors lies in the computing power of their products and the high-precision maps produced by crowdsourcing.

Delphi and Mobileye announced in late August that they will jointly develop a complete set of SAE (American Society of Automotive Engineers) Level 4/5 autonomous driving solutions. This end-to-end, mass-producible, high-performance and safe fully autonomous driving solution can be quickly integrated into a variety of automotive platforms of different customers around the world. The CSLP platform jointly developed by the two parties will be demonstrated at the 2017 Consumer Electronics Show in Las Vegas, USA, through mixed sections of cities and highways, and is planned to have mass production capacity in 2019. Last week, the two parties said that Intel will provide customized chips for the platform, which can achieve 20 trillion mathematical operations per second, and the updated version planned to be launched later is expected to have two to three times the computing power of the first generation of products.

Last Tuesday, the media was invited to test drive the Audi SQ5 self-driving car modified by Delphi in Pittsburgh for a 10-minute journey, including several uphills, downhills, curves and parking intersections. An engineer sat in the driver's seat, but he did not touch the steering wheel during the entire process. The test ride experience was nothing special, similar to most self-driving trips. The control system is very cautious, and the car is programmed to drive automatically below the average speed to ensure a safe driving path.

However, the Audi SUV equipped with Delphi's self-driving system that the media took a test ride in looks very different from the self-driving test cars of Google, Ford and Uber. First of all, there is no lidar installed on the roof of the car, and there is no visible camera. Secondly, there are no special sensors installed on the exterior of the car. Glen DeVos, vice president of Delphi Engineering and Services, told reporters that their goal is to build a driverless car that looks no different from the vehicles on the road today.

The entire test ride lasted about 10 minutes, starting from a parking lot of Ottomatika, driving along a main road, passing through several office areas, and then returning along the original route. Although no human intervention was required during the entire driving process, the driver still sat behind the steering wheel to ensure driving safety.

Note: Ottomatika is an emerging company researching autonomous driving and was acquired by Delphi last year. Ottomatika was formerly a company established by Carnegie Mellon University to study how to commercialize autonomous driving technology.

During the test drive, the vehicle stopped before reaching a stop sign. The car moved forward a little, but then stopped again. DeVos said this is exactly the part they want to debug. In this case, the vehicle's autonomous driving system will make a judgment. Usually, the system will choose a safe driving method. If the system finds something uncertain, it will stop the car first.

However, what makes Delphi and Mobileye so confident that they can sell self-driving cars in the next two years? The company's head said that the high-performance computing power provided by Intel chips is jointly developed with the core technologies of Mobileye and Delphi. Mobileye will provide EyeQ4/5 system chip technology with sensor signal processing, fusion and world view generation functions, as well as Road Experience Management (REMTM) that can achieve real-time mapping and vehicle positioning functions. Delphi will provide the autonomous driving software algorithm with road and motion trajectory planning functions from the acquired Ottomatika company, as well as the Multi Domain Controller (MDC) that integrates a complete set of camera, radar and lidar systems.

In addition, the two parties will also work together to develop the next generation of sensor fusion technology and formulate the next generation of simulated human "driving behavior specifications", integrating Ottomatika's driving behavior modeling system and Mobileye's deep reinforcement learning system, so that autonomous vehicles can communicate and negotiate with other human drivers and pedestrians in complex urban driving environments.

Unlike other companies that rely on a fleet of driverless cars to collect and map specific cities, Delphi and Mobileye "crowdsource" the mapping work to tens of millions of traditional vehicles equipped with Mobileye's vision technology.

Mobileye said that there are more than 14 million cars on the road equipped with the company's front- and rear-mounted ADAS products. The data captured by these cars can be transmitted to self-driving cars to help them better predict real-time road conditions, such as construction sites and double-parked cars.

Dan Galves, senior vice president of Mobileye, said that using ordinary cars driven by humans to create maps is of great benefit. While the self-driving car is driving, it can ask the system, "Dear map, please tell me where the thousands of other cars are driving."

Mobileye calls this technology "Road Experience Management (REM)". This technology can provide real-time data to continuously update and optimize electronic maps to help autonomous vehicles navigate. Unlike other electronic maps with huge amounts of data, this type of data is easier for vehicles to maintain and store, and this technology can work without wireless connections.

Of course, Delphi and Mobileye are not the only companies that use crowdsourced maps to provide navigation for self-driving vehicles. In July this year, Ford invested in a company called Civil Maps. Unlike Mobileye, which only relies on visual sensors, this company uses data obtained from multiple sensor devices to draw maps. Google is also using this crowdsourced mapping method. Sam Abuelsamid, a senior analyst at Navigant, a consulting company, said that almost every company researching autonomous driving is now moving in the direction of crowdsourced maps.

Abuelsamid pointed out that in terms of corporate revenue, Delphi is far ahead of large companies such as Tesla and Uber, and Delphi has maintained a supply relationship with many OEMs. As an experienced parts supplier, Delphi's advantage lies in its understanding of how to design, verify, produce and support autonomous driving rating systems. In addition, Delphi can also provide complete advanced driver assistance systems and autonomous driving systems for automakers that do not have the resources to develop their own autonomous driving technology.

In July this year, BMW announced a partnership with Mobileye and Intel to launch a mass-produced version of driverless cars in 2021. In addition, Mobileye's EyeQ series vision SoCs have also been adopted by General Motors and Volkswagen. Just in August, Delphi launched a batch of autonomous driving test cars in Singapore. These autonomous driving cars were tested to cater to Singapore's Mobility-on-Demand Program.

It is undeniable that Mobileye and Delphi are also facing huge challenges and competitive pressure from all sides. Chipmaker Nvidia's Drive PX 2 autonomous driving platform has been adopted by Audi and Tesla. It has stronger computing performance and the GPU itself is more suitable for running deep learning algorithms. This has also brought considerable pressure and challenges to Intel, which wants to recreate its glory in the field of automotive chips.

Delphi and Mobileye said they will demonstrate their latest self-driving technology at the 2017 CES International Consumer Electronics Show. The vehicle equipped with this unmanned driving technology will start from the core of Las Vegas and travel about 6.2 kilometers, passing through highway merges, passing through crowded urban roads full of pedestrians and bicycles, and passing through tunnels.

Compared to Uber, which has just entered the field of autonomous driving, Delphi and Mobileye have been researching autonomous driving for a much longer time. This driverless experience was demonstrated through their prototype vehicle, which has a clean appearance and you would not realize that it is a car that can drive itself. This experience is also the uniqueness of the Delphi/Mobileye autonomous driving solution.

As a winner of Toutiao's Qingyun Plan and Baijiahao's Bai+ Plan, the 2019 Baidu Digital Author of the Year, the Baijiahao's Most Popular Author in the Technology Field, the 2019 Sogou Technology and Culture Author, and the 2021 Baijiahao Quarterly Influential Creator, he has won many awards, including the 2013 Sohu Best Industry Media Person, the 2015 China New Media Entrepreneurship Competition Beijing Third Place, the 2015 Guangmang Experience Award, the 2015 China New Media Entrepreneurship Competition Finals Third Place, and the 2018 Baidu Dynamic Annual Powerful Celebrity.