TU-TESL

• 22

In order to better promote cross-border collaboration in the global scope, clarify the roles of travel companies, automobile companies, cities and other entities in the future travel ecology, discuss the major changes that will occur in the future travel mode, travel technology, traffic structure, traffic strategy and the opportunities and challenges that the whole industry will face, China electric vehicle 100 people's Congress organized 2019 The global future travel conference aims to promote the ecological transformation, transformation and innovation of travel and international coordination to meet the travel revolution. The following is the speech of Yang Liangyi, an expert in intelligent vehicle test and evaluation of China Automotive Engineering Research Institute Co., Ltd.: Thank you for your introduction. Good afternoon, everyone! I'm Yang Liangyi from China Automotive Engineering Research Institute. I'm very glad to have this opportunity to share with you some of the work and some thoughts our team has done in the field of test and evaluation of ICV. Today, the topic of my speech is the construction of the test and evaluation system of ICV based on China's traffic characteristics. It is mainly divided into four parts. First of all, briefly describe the background and basic framework of our system construction. Traffic safety, energy consumption and traffic congestion are common problems faced by traffic. At present, cars account for 33% of oil consumption. More than 90% of traffic accidents are caused by human factors, and there are many reasons for traffic congestion. Irregular driving behavior is one of the reasons. Based on such a traffic situation, ICV has a good technical means and platform to solve the above traffic problems. At present, this area has also become a key development strategy of our country. When it comes to this, why should we mention the characteristics of China's traffic environment? As we all know, many sensing modules in the intelligent network vehicle mainly include radar, camera head and positioning system. In fact, the main perceptual information of these perceptual modules is actually the traffic environment information. So the intelligent Internet car is inseparable from the traffic. In fact, China has a vast territory. From the south to the north, from the east to the west, the road types are very complex and the climate is quite diverse. In many urban roads, the phenomenon of people and vehicles mixed is very common. There are a lot of people and vehicles who don't follow the traffic habits in the time of congestion. This is a typical feature of China's traffic environment. At present, intelligent vehicle testing methods mainly from three aspects, one is simulation testing, in the algorithm and extremely dangerous scenario testing. The second is the closed field test, which mainly solves the key technologies of intelligent vehicles. The third is the open road side test, which is the comprehensive verification of the whole vehicle system, mainly the road adaptability. At present, in terms of these links, there are still some R & D testing links in the standard laws and regulations, mainly including closed field testing and a small number of road side tests. This is more than L3 test, and the proportion of open road test and simulation test will be higher and higher. This is the development trend of testing. When it comes to China's traffic characteristics, what are the main connotations of China's traffic characteristics? The main problem of traffic characteristics in China comes from some data of natural driving and traffic accident data. Based on this research, we form the scene database of traffic scenes in China. In natural driving, there is a study on the characteristics of driver behavior, which can form a model of driver behavior characteristics. There are also some key parameter sources for product development. At the same time, China's traffic scene database and typical driver behavior database can be solidified into various test scenarios, which can run through all aspects of the test. Including vehicle model benchmarking and the formulation and evaluation of final standards and regulations. The whole construction of Vista is divided into six parts, with the analysis of key parameters of Chinese typical scene database and typical vehicles as the beginning of the part. The key test and evaluation of the typical model of mature mass production model, index evaluation as the last link in the whole process. We apply typical traffic scenarios and important parameters of China to typical test scenarios. The whole scene also runs through the six links of the system. Combined with the special test equipment including the traffic monitoring center, a complete test evaluation system is formed. Second, I'd like to briefly introduce some of the work i-vista has done. China's traffic scene source, natural driving data collection. At present, we have carried out large-scale natural driving data collection in the country. We have run 29 provinces and collected 600000 kilometers of data. About 100 drivers were replaced in the middle, and their responses were analyzed to form a driver's characteristic model. This is what we do in the area of data collection. We need to transform the collected data of natural driving into scenes, and there is a series of work to be done in the middle. I-vista combines its own research tools and general software tools to build the whole scene data processing tool chain. Combined with our own natural driving scene, we have formed a rich test scene. The test scenario is then superimposed with weather conditions, traffic factors and road factors to form a typical Chinese traffic scenario database. In October 2018, we released the first version of China's typical driving scenarios in the industry, including four main scenarios. Dangerous scene and parking scene, lane changing scene and following scene. This year, we released the second version of the scene, which is rich in content. I hope you can pay attention to it. In order to verify the adaptability and performance of the scene, we combine the scene with the hardware in the loop test system. We cooperate with internationally renowned simulation tools to support the testing and verification of video and millimeter wave radar in the ring. This part is now carrying out the corresponding controller test work. The third part is about our work in vehicle test evaluation. Vehicle evaluation mainly includes two dimensions: objective evaluation and subjective evaluation. At present, many tests are mainly based on objective evaluation, supplemented by subjective evaluation. Objective evaluation mainly includes in-depth benchmarking of typical models, compliance verification of standards and regulations, and R & D in the process of product development. Vehicle evaluation is mainly such a framework. For this reason, a large number of targets have been purchased, including 360 degree fake vehicles, high-precision rapid mining and monitoring center. These societies form the special equipment for i-vista site test. We have carried on the transformation in the traditional experimental field, for example, we have added many typical curve scenes. We have also set up a new type of experimental base for intelligent Internet connected vehicles in Chongqing. For the test of higher-level intelligent driving, at present, two test areas of automatic driving are promoted. In the future, there will be more areas for open road test of high-level intelligent driving. Based on the closed field test, we focus on the verification of intelligent vehicle performance and functional logic. At present, we have formed a relatively complete closed field test capability, including the high-level intelligent driving test capability of L2 tgicc. Based on the open road, we have established a complete road monitoring and testing system, which can carry out the evaluation and verification of road adaptability of intelligent vehicles. In the research of standards and regulations, we have also established a perfect research system of standards and regulations, which mainly includes the authorized standard capabilities, reaching more than 30. It includes the active safety testing capabilities of Euro NCAP, the most difficult in the world. The inspection of the whole i-vista standard and regulations is very perfect. For the evaluation of mass production models, when consumers choose to buy a car, there is a lack of some reference for selection. Now many enterprises are faced with the problem that the direction and index of product optimization and improvement are missing in the process of product development. I-vista has established the China intelligent vehicle index to evaluate the performance and function of current mass production models. In 2018, the first version of the formal procedures was released, 25 mass production models were evaluated, and the voice continued to be heard on the official website, which was publicized by the media. In September this year, the second version of the regulations was released, adding L2 intelligent driving. In the last part, in order to improve the test efficiency and pass on our test experience and understanding, we infuse these things into the development of special test equipment. We have made a special part to develop the equipment. Our advantage lies in that we are originally the makers of standards and regulations and also the implementation units of standards and regulations testing. At the same time, we provide very rich testing services to the industry, and we are also the users of many very well-known equipment in the world. We put forward a testing equipment development system, which takes testing as the center, the problems found in testing as the core, and the simplicity and easy operation of the equipment as our focus. At present, the first generation of products are mainly for side test of open road. We have developed a set of equipment, mainly combined with the traffic monitoring system can complete the test and evaluation of the open road. Our i-tester equipment can be combined with high-precision maps and other equipment to form a complete set of solutions for closed field testing. The post-processing software tool of our experimental data just mentioned in this part is mainly based on the test experience and found that many devices are used and there are many inconveniences in data analysis. We developed a set of post-processing software, which can improve the data efficiency in the experiment process and greatly improve the convenience in the test part. Our tools will continue to work in this area. In the future, we will push back the integration of testing and tool services. This is our main direction in the future, which can provide users with a test plus tool integrated service solution. Testing and tools are areas and sectors that we continue to work on. We hope more partners join us, do more cooperation, and do more things that are conducive to the development of the industry. Thank you!