“From laboratories to test sites to mass-produced cars to the formation of stimulating consumption, how smart cars can transform from assisted driving to autonomous driving technology, and how are automotive electronics companies moving from traditional embedded electronics companies to AI-level embedded platforms Transitional?
Smart cars have mastered the technology of sensing the environment inside and outside the car, analyzing and planning driving routes, and automatically controlling driving, and gradually gained the ability to intelligently assist human driving. With the gradual breakthrough of technical difficulties, Advanced Driver Assistance Systems (ADAS, Advanced Driver Assistance Systems) is gradually transitioning to autonomous driving.
From laboratories to test sites to mass-produced cars to the formation of stimulating consumption, how smart cars can transform from assisted driving to autonomous driving technology, and how are automotive electronics companies moving from traditional embedded electronics companies to AI-level embedded platforms Transitional?
On November 13, the “OFweek 2018 (3rd) China Artificial Intelligence Industry Conference-Smart Cars, hosted by OFweek, China’s high-tech industry portals In the special session, “As a manufacturer of electronics, VIA Electronics is currently transforming to become a provider of autonomous driving solutions.” Qin Shu, VIA Electronics Marketing Manager, discussed the key technological innovations from assisted driving to autonomous driving in his speech, and introduced Some successful experiences of VIA Electronics in the process of making autonomous driving solutions.
(Qin Shu, VIA Electronics Marketing Manager)
VIA Embedded Electronics’ transition to an AI-level platform
As an integrated circuit design company in Taiwan, VIA Electronics mainly produces motherboard chipsets, central processing units (CPU), and memory. It is the world’s largest independent motherboard chipset design company. Qin Shu introduced that VIA’s transformation path started from the acquisition and gradually progressed to the embedded platform.
On the road to the transformation of autonomous driving, Siegwell Electronics is also making a step-by-step transition from assisted driving to autonomous driving. Driving assistance systems include lane keeping assist systems, automatic parking assist systems, brake assist systems, reversing assist systems, and driving assist systems.
“With the help of artificial intelligence and computer vision to empower the embedded platform, the front camera solution is currently its main sensor choice for assisted driving.” Qin Shu introduced that the company’s current ADAS functions include: lane departure warning, 3D surround view, and driving record , Blind spot monitoring, road traffic prediction, etc.
In his speech, Qin Shu analyzed the basic situation of China’s new energy bus market in 2017. China regards new energy vehicles as one of the seven strategic industries. Encouraged by many positive policies, China’s new energy vehicle market demand spirals upward. , The total demand for new energy vehicles in 2020 is planned to be 2 million vehicles. At the same time, with the rapid development of China’s new energy vehicles, the Chinese government is beginning to study new incentive policies to promote the development of China’s new energy vehicles in the direction of intelligent networked vehicles. For example, Qin Shu mentioned that with pure electric vehicles as the main breakthrough point, VIA Electronics is working hard on public transportation vehicles!
The VIA Mobile360 intelligent driving assistance solution, as one of the suppliers of intelligent driving modules in the new energy driverless bus line released by Huzhou Enchi Automobile, provides the VIA Mobile360 SVS panoramic surround view vehicle body monitoring technology; equipped with VIA Mobile360 ADAS advanced Driving assistance system; through the car FARKA connector, connect 5 independent high-definition cameras, which can support up to 8 cameras; hardware performance includes: rugged host, wide temperature, wide pressure, anti-shake and anti-shake, designed for all kinds of harsh vehicles Environmental design; software functions include: 360-degree panoramic view of body monitoring, lane departure and collision warning, blind spot speed limit monitoring, driving record, fleet remote monitoring and tracking; software and hardware can be flexibly customized, suitable for various harsh road and vehicle environments.
“China’s domestically produced new energy vehicle industry is developing rapidly. Pure electric buses are one of the most important strategic deployments. The demand for intelligent development cannot be underestimated,” said Qin Shu. “The VIA Mobile360 intelligent driving assistance solution is a set of The software and hardware package solution that combines all-round body monitoring and ADAS can be quickly deployed in the vehicle environment to help customers seize the opportunities in the intelligent new energy vehicle market.”
How many steps are still missing from assisted driving to autonomous driving?
From assisted driving to highly automated driving to unmanned driving, the work of mapping and positioning must be done first. How can we do a better map and positioning? Qin Shu believes that in the traditional sense, a map is a navigation map, with road information, intersection information, some traffic rules information, etc.; but to achieve a high degree of autonomous driving, more information is needed, such as traffic signals and the number of lanes. , Slope and curvature, etc.; and for unmanned driving, high-precision maps may be required. This includes more detailed modeling of the environment. Various road signs and road semantics are reflected, and at the same time, it is also required Give some driving advice. For example, where do you need to slow down, where do you need to change lanes…
In addition to positioning, better cognitive algorithms are needed. Autonomous driving is divided into three stages: perception, planning, and control. In the advanced stage of perception and part of planning, better cognitive algorithms are needed. From the perspective of advanced perception, first of all, traditional assisted driving only needs to recognize specific goals. For example, on high speeds, you only need to recognize cars, and at some intersections, you need to recognize pedestrians. Highly autonomous driving may drive into many very complicated working conditions, and there is no way to have a limited database for cognition.
Traditional vision solutions require a database, and highly automated driving requires more complex modeling of the world. Qin Shu finally sighed that VIA Electronics will continue to explore the deep learning algorithms of smart cars, adding a force to the development and technological innovation of smart cars!