What is the biggest problem facing the development of autonomous vehicles? Different fields can give their own different views.
As an emerging thing, compared to traditional cars, autonomous vehicles not only involve a lot of the latest technology, but also in software development, they are even better than drones or aerospace shuttles. , The problems that need to be dealt with are also more complicated.
As the first step in the design of autonomous vehicles, simulation plays a vital role in simplifying the design process, shortening the design time, and verifying the design results.
However, like Other self-driving car development processes, simulation also faces many problems in the development and testing process. For example, how to design self-driving cars to best ensure its reliability can be obtained in the simulation stage. The solution.
Speeding up the resolution of this type of problem is a problem that MathWorks emphasized at this year's MathWorks China Automotive Annual Conference.
Lin Xiaocang, Director of Embedded Coder Product Family Development
Two digital revolutions in car developmentTo a certain extent, the development of autonomous vehicles is a process in the digitalization process of traditional car development. It is the only way to digitalization, and it is also an inevitable result.
In the view of Lin Xiaocang, the development director of MathWorks Embedded Coder product series, this result is not achieved overnight. From a development perspective, there are at least two steps.
The first step, and also the initial digital transformation, refers to the ubiquitous penetration of embedded software into traditional car development. I still remember the old cars that used to have no airbags, no ABS, and no music. Some are just the most primitive functions of the car, driving, and the development process is relatively simple.
With the addition of more and more functions, the application of embedded software is imperative. Car development is based on driving and starts to consider fuel economy and emissions, electrification and vehicle safety, comfort and convenience, etc. Various problems.
With so many problems to be solved urgently, it is natural not to wait until the prototype is produced to discover and solve the problems. Starting from model design, we can design a car that keeps pace with the times faster.
"At this time, the car model design generally goes through three cyclical steps of modeling and simulation, testing and verification, and code generation."
The second digital revolution is that with the advent of autonomous vehicles, artificial intelligence has become ubiquitous.
Lin Xiaocang emphasized: "Data-driven algorithms, machine learning, deep learning, and autonomous systems have penetrated into the positioning, planning, control, and perception system modules of autonomous vehicles, and have also made the simulation tools required by autonomous vehicles more complicated. "
One problem brought about by the increase in tool complexity is that the difficulty of getting started has increased exponentially. Since autonomous driving involves a wide variety of fields and technologies, they are divided into categories. It requires that the knowledge that the developer knows must be all-encompassing, even reaching the point of omniscience. For developers, the requirements can be said to be sensational.
However, after all, there are specialties in the technical industry. Developers are only proficient in a certain field. They are not necessarily specialized programmers, and they do not necessarily have very rich knowledge and background in other fields. They have to rely on computer programs just because of the needs of the product. To deal with tasks that have never been deeply understood, how to solve these problems better and faster? Lin Xiaocang believes that at this time, what developers need most is modular, highly integrated tools that can make up for developers' shortcomings.
What tools can make up for developers' shortcomings?It can be said that model-based design has been widely used in the development process of various industries, not only in the development of autonomous vehicles.
The advantage of this design is that for developers, they can quickly apply knowledge in other fields without requiring too deep knowledge, which virtually speeds up the design and simplifies the design process. Therefore, in more and more development processes, model-based design plays a key role in today's highly autonomous systems.
Lin Xiaocang said that MathWorks hopes to provide developers with a series of design and verification tools through the introduction of new tools, so as to realize the rapid development of artificial intelligence algorithms and simplify the verification of autonomous vehicle algorithms.
To this end, in 2017, MathWorks launched the Automated Driving System Toolbox (ADST), hoping to provide a series of tools to accelerate the development of autonomous driving systems and active safety systems for engineers engaged in autonomous driving and active safety system design.
It is understood that this tool mainly includes three aspects:
First, testing and verification tools, including true value labels and scene generation;
Second, algorithm development tools, including sensor fusion and target tracking, and vision system design tools;
Third, visualization tools, including bird's eye view tools, and radar point cloud tools.
As we all know, autonomous vehicles expand the concept of the environment more broadly, including weather, traffic, and road conditions.
ADAS and smart cars include a lot of environmental sensors, including cameras and various radars with different functions, such as ultrasonic radar, millimeter wave radar, and lidar. These all pose great challenges to modeling and simulation technology.
MathWorks provides complete modeling tools and methods for the above functional applications. Scene working conditions, sensor fusion, control algorithm, actuator, these four parts are a complete intelligent driving simulation architecture, this tool can be solved one by one.
It is understood that in the field of core algorithm development software tools, MathWorks almost occupies a monopoly position in the global market.
Mobileye, Delphi, etc. all use their software tools to develop modules such as camera, millimeter wave radar, and lidar. Moreover, almost all OEMs also use its products for implementation and control module development.
In addition, Lin Xiaocang also emphasized that in each stage of model-based design and development, MathWorks has different toolboxes to support rapid, efficient, and high-quality completion of the corresponding development work, and a professional training team provides standardized and customized training courses.
"For developers, how to design faster and better is the most important thing. Among them, a suitable tool plays a very important role, and the same is true for the development of autonomous vehicles." Lin Xiaocang finally said, "As an emerging thing, self-driving cars will encounter problems that have not been encountered before in the development. We are very willing to cooperate with more manufacturers to gradually solve the problems in the self-driving system. Make a design that can really accelerate, even Tools to complement the shortcomings of developers!"
Ecome mobile radio,Portable Two Way Radio,Motorola Vhf Handheld Radio,Multi Band Portable Radio
Guangzhou Etmy Technology Co., Ltd. , https://www.gzdigitaltalkie.com