Since Tesla unveiled the scenes of automobile intelligence, electrification and intelligence have begun to accelerate the penetration of the automobile market. Especially since 2021, the new domestic car-making forces represented by Xiaopeng, Weilai, Jidu, etc., have all taken accelerating the process of product intelligence as a strategic task for enterprise development.
This is true at home, but also internationally.
Looking around the global auto market, in order to seize the opportunity of the tide of automobile intelligence, not only are there "new forces" successively announcing the official construction of cars, but also countless traditional car manufacturers continue to increase their intelligent exploration: at the 2022 CES exhibition, Sony launched the SUV prototype VISION-S02, and announced its entry into the field of intelligent electric vehicles; Volvo, Cadillac and other traditional car companies that are at the forefront of intelligence are also continuing to increase their layout in the field of intelligence, and BMW has launched a one-time launch including "BMW suspension giant screen, The four major technologies, including the new My Modes theme mode, the BMW electric vehicle sound wave simulation, and the appearance of the color-changing car based on electronic ink technology, can be called an intelligent "gift package".
The question arises, in the dazzling concept and function demonstration of smart cars, what is the most important measure of smart cars?
The answer is just as Jidu CEO Xia Yiping said in a recent interview that "hidden under the sea level, the real 'fearful' iceberg picture", the "iceberg" problem that is not easy to see is often the most important problem to be solved.
The safety and stability of the intelligent system of the car, including intelligent driving and smart cabin, is the "iceberg" of the intelligent car.
The battle over autonomous driving routes may not be conclusive, but software security issues have never had one answer
Just recently, Mercedes-Benz announced that when drivers equipped with Drive Pilot open the vehicle's advanced driver assistance system, Mercedes-Benz will bear legal responsibility for the operation of the car.
In fact, the regulation stems from the German government's mandatory requirement for the release of conditional autonomous driving at L3 level. The commitment is limited to low-speed road conditions in some congested scenarios and cannot cover the commonly used speed range, so it is considered by some people to be "gimmicks are greater than reality". The legal liability borne by Mercedes-Benz also has more restrictions, such as high-precision maps, electronic fences and so on.
Safety is a systemic issue, and the division of responsibilities is only a small part of it. In the era of smart cars, how to make cars safer is the key, and software security is the top priority. The division of responsibilities between OEMs and consumers is ultimately unconfident in the safety of autonomous driving software.
The safety of autonomous driving has always been a route dispute: a gradual route that starts with L2 and then develops to advanced automatic driving; and a leapfrog route that directly develops L4 and realizes true automatic driving.
At present, compared with L2 level automatic driving, the performance of L4 autonomous driving of autonomous driving technology companies is better. In addition to a variety of publicly available all-unmanned services and activities, there is also data to prove that the Department of Motor Vehicles released data showing that there was not a single fatality in the 149 autonomous driving accidents from 2019 to 2020. The "automatic driving leads to accidents" that we see every day is basically drivers who over-trust L2-level automatic driving.
Comparing the invisible confrontation between OEMs and autonomous driving technology companies, it is not difficult to find that in the wave of autonomous driving industrialization, although autonomous driving companies are far from the public, they are at the forefront of absolutely safe automatic driving with fully unmanned driving practice. Google Waymo and Baidu Apollo are relatively far away from ordinary passenger car users, but when it comes to the safety of autonomous driving, we can't avoid discussing them.
Why is the "Waymo, Apollo" autonomous driving safer?
Not only the practice of road testing, in the industry and academia, "Waymo, Apollo" is safer than car companies that conduct automatic driving research and development based on L2, and have become the mainstream cognition.
In the "White Paper on Traffic Safety of Autonomous Vehicles" jointly released by China Automobile Center, Tongji University and Baidu at the end of last year, through the comparative analysis of authoritative technical demonstrations and real autonomous driving accidents, not only the conclusion that "autonomous vehicles can effectively avoid 80% of the accidents of human driving when driving normally", but also revealed the operation mechanism of Apollo's high-level automatic driving safety through a large number of professional analysis.
As the only Chinese company to rank in the international autonomous driving "leader" camp for two consecutive years, Baidu can be called the leader of China's autonomous driving. We may wish to take Baidu automatic driving as an example, through in-depth interpretation of safety, reveal the secret of its high safety of automatic driving.
The first is the design of the three-layer safety system for automatic driving. Up to now, Baidu Apollo has formed a three-layer safety system of main system safety, redundant safety system and remote cloud driving system, which comprehensively ensures Baidu's L4 level automatic driving safety:
Through the safety of the main system, it is possible to make the autonomous vehicle reflect the speed higher than the human driver and ensure that the vehicle behaves in accordance with the right of way;
The safety monitoring and redundancy system can realize full redundancy of hardware and vehicles, even if any single component of the kit fails, it will not affect safety;
Using 5G cloud driving, you can "one-to-many" control of autonomous vehicles running on normal roads and guard the last line of safety defense.
In addition, through simulation-driven rapid verification, Apollo can fix more than 90% of new collision problems every six months, and complete the test and verification system for unmanned people.
Under the premise of doing a good job in safety systems, Apollo has obtained the world's leading large-scale high-level landing practice data accumulation, as of March 2022, Baidu Apollo's latest progress is: road test mileage of more than 25 million kilometers; 411 Chinese autonomous driving licenses.
One piece of data that is difficult for other companies to do is the real-world road L4 autonomous driving test that Baidu's Apollo road test fleet conducts non-stop in more than 30 cities across the country. Apollo's 25 million kilometers of autonomous driving road test mileage will continuously feed back the road test data to the algorithm, improve the Cornerstone (corner case), and make the automatic driving system safer.
How can high-level autonomous driving empower mass production of smart cars to be safe?
Since the autonomous driving of "Waymo and Apollo" is so safe, how should they use their advantages at this stage to empower the safety of mass-produced smart cars?
Jidu automotive robots have begun to practice.
As the only intelligent car brand in the industry that applies Baidu Apollo's high-level automatic driving capabilities in the full-stack application, Jidu Auto Robot inherits the full set of capabilities and safety systems of Apollo's high-order automatic driving, and can be called the "seed player" of mass production of high-level automatic driving.
As Jidu CEO Xia Yiping said in a recent public speech, when smart cars enter the 3.0 era, the requirements for software security in automobiles have also been correspondingly improved. Therefore, the research and development of software, the verification of software capabilities, and the security and stability of software operation have become the core of Jidu's most concerned in the era of intelligent car 3.0.
In order to achieve the goal of software security, Jidu has developed its own electrical and electronic architecture and domain controller, from the perspective of combining software and hardware, to fully ensure the security of the overall architecture; through the fusion perception of lidar and vision solutions, improve the hardware redundancy of perception; and configure the NVIDIA Orin chip dedicated to intelligent driving, which has higher computing power redundancy. Through the development logic of software and hardware, Jidu has improved the certainty of the safe landing of high-level automatic driving capabilities.
In the research and development process, Jidu has also created a set of working methods exclusive to the "brain" of automotive robots: through the "SIMUCar software integration simulation prototype", pre-intelligent development, with the optimization of the production process to improve product safety, and the cycle of safety verification is lengthened. Through the dual-track parallel of the decoupling of the development of the vehicle's software and hardware, Jidu put forward the research and development goal of "software first to improve safety", indicating that without relying on OTAs, the factory can fully have the corresponding automatic driving capabilities. Jidu's autopilot technology for Apollo was developed and tested again on the basis of its atomization capabilities.
At the end of last year, the development of high-end automatic driving functions has achieved high-speed, urban domain dual-domain integration, and after the SIMUCar upgrade version 2.0, it has completed the integration of high-level automatic driving software and vehicle electronic and electrical architecture and SOA.
postscript:
At the end of Qingping, while the main engine factory and the automatic driving technology company are constantly moving, governments are also actively promoting the development of new energy and intelligent vehicle related industries. Just on March 22, the National Development and Reform Commission and the National Energy Administration issued the "14th Five-Year Plan for Modern Energy System", according to the plan, by 2025, the sales volume of new energy vehicles in mainland China will account for about 20%.
In fact, this judgment coincides with the judgment of a number of industry leaders:
Baidu founder Robin Li believes in his new book "Intelligent Transportation: Major Changes Affecting Mankind in the Next 10-40 Years" that "in another decade, by about 2025, the vast majority of cars running on Chang'an Avenue should be autonomous vehicles with intelligent, automated, networked and electric vehicle characteristics", and Weilai founder Li Bin also has the famous remark that "2025 will be the year of the electric vehicle showdown".
Will 2025 really be a key year for smart electric vehicle competition? We don't know about this, but perhaps Xia Yiping's sentence can be answered: "What is more difficult is the things under the iceberg, we must do a good job under the iceberg, don't just focus on the things on the iceberg, and the basic work cannot be lazy." No matter which year in the future will become the decisive year of smart electric vehicles, anxin will work the software running performance that the outside world cannot see, which is the greatest reverence for the product and the market.