For a long time, autonomous driving has hovered between accident and safety.
On the evening of March 18 this year, an Uber self-driving car killed a pedestrian crossing a road in Tempe, Arizona. The victim was a 49-year-old woman. After investigation, Uber stopped the automatic driving test because of the driver's inattention and failure of the automatic driving system.
Almost at the same time, a man in California drove a Model X into a highway barrier. At that time, the vehicle was in "auto-driving state" and caught fire after the impact. The driver died after being taken to the hospital.
Tesla’s accident was not the same, which indirectly led to Tesla withdrawing the long-advertised "fully autonomous driving" option from the order page of the company's website this month. Musk said: "Fully automatic driving" option is being prepared by the company. Before the launch of the fully automatic driving function, it will no longer appear in the menu, but it will be temporarily provided "out of the menu".
Musk believes that the fully automated driving option will “create too much confusion†for customers and cannot guarantee that the driver will always pay attention to the road and take over in time.
Hong Kong Applied Science Research Institute Yang Meiji said: Autonomous driving is limited by perception, deep learning, processor performance and other factors. The technology is still difficult to mature in the short term. High-speed communication connection technology can make up for the lack of intelligence on the end.
Yang Meijian became the Chief Technology Officer of ASTRI in January 2016. Before joining ASTRI, he held senior management positions in Taiwan Semiconductor Manufacturing Company and IBM Research Institute.
Dr. Yang holds a PhD in Electrical and Electronic Engineering from the Massachusetts University of America and an MBA from the MIT Sloan School of Management.
Focus on scientific research, close to public welfare
Hong Kong Applied Science and Technology Research Institute Co., Ltd. (ASTRI) was established by the Hong Kong Special Administrative Region government in 2000. ASTRI’s main research and development fields include communication technology, electronic components, hybrid signal system chips, advanced digital systems, optoelectronics, and information security. It is composed of seven technical departments of data science and intelligent software and systems.
Yang Meiji said that ASTRI focuses on the application of scientific research and technology. At the same time, because it is established by a government organization, the commercial atmosphere is not strong, and public welfare and scientific research credibility are dominant. This is why many companies are willing to cooperate with them.
In terms of visual recognition and deep learning, the institute has conducted explorations and trained different neural networks. The existing neural network's recognition of targets is still in the stage of continuous development and upgrading.
In the laboratory, the trained model has a good effect on target recognition, but in practical applications, it is often necessary to perform different simplifications for subdivision areas. The purpose is to ensure that the recognition effect will not change much when the network model is reduced by multiples.
Especially on the embedded side, low power consumption and low latency are more valued. ASTRI has also conducted research and experiments in the field of autonomous driving. Autonomous driving requires a stronger processor to run a deep learning network, but at the same time it has a low-power, high-calculation processor.
In order to make the deep learning network performance better, the chip-based deep learning algorithm is an industry consensus. Yang Meiji believes that FPGA and ASIC are two directions at the moment.
However, due to the continuous changes of deep learning networks, FPGAs can more easily adapt to flexible and changeable algorithms. Therefore, the current deep learning applications choose FPGAs more frequently.
More optimistic about C-V2X
Since this is true, the current technology of autonomous driving is still far from being mature, so ASTRI has explored in another direction-the new generation communication network (NGN).
Yang Meiji said that ASTRI has already conducted in-depth cooperation with the mainland government and enterprises, such as a demonstration of wireless communication in Wuxi, Jiangsu Province with Huawei.
By installing communication equipment on both sides of the road, the car can predict the situation of the road beyond the visual range and the situation of pedestrians passing by the intersection in advance. Through the communication between the roadbed equipment and the vehicle, the lack of perception and intelligence of the autonomous driving end is solved.
Next, ASTRI will also demonstrate some car networking applications in Europe with Huawei.
Compared with the intelligent improvement of the autonomous driving end, relying on C-V2X can help vehicles perceive traffic information more accurately, and the realization path of this method is obviously faster and more realistic than the former. Therefore, Yang Meiji has great expectations for the development of the C-V2X industry in the next few years.
The foundation of the development of C-V2X Internet of Vehicles is the high-speed communication network. At present, the industry takes 5G as the direction. Due to the particularity of the automobile industry, the requirements for network communication are strict, and the main performance indicators are low latency and high traffic.
Around 4G / 5G base stations, drive test terminals and vehicle terminals, there is a lot of work to be done to solve this goal. ASTRI is committed to research, development and commercialization of 5G and 5G wireless systems' cutting-edge communication technologies and solutions, hoping to solve 5G communication problems.
Randm Vape,Randm 10000Puffs Disposable Vape,Randm Disposable Pen,Randm Disposable E-Cigarette Pen
Shenzhen Uscool Technology Co., Ltd , https://www.uscoolvape.com