Establishing Norms, Standards, and Data to Safeguard Unmanned Safety

Driverless, as the name suggests, can automatically handle all kinds of road conditions and emergencies without the intervention of a driver. In spite of the rapid advancement of unmanned technology in recent years, it has not reached the same level of intelligence as humans. In particular, it has not been able to weigh the pros and cons as humans have done in terms of trade-offs before the accident. How to ensure unmanned safety has received wide public attention. This issue of this issue will analyze driverless safety issues in terms of product technology and social ethics.

In the past two years, driverlessness has become a hot spot in the field of artificial intelligence. Many technology companies including Google and Uber, as well as traditional car companies such as Ford and Toyota, are all launching their own driverless technologies. In China, Baidu, Tsinghua University and other universities and institutions of higher learning have also invested heavily in driverless technology and achieved substantial results. The Hongqi HQ3 unmanned vehicle independently developed by the National University of Defense Technology has for the first time completed a full-speed unmanned pilot test of 286 kilometers from Changsha to Wuhan.

Gong Jianwei, associate professor of vehicle engineering at the Beijing Institute of Technology School of Mechanical and Vehicle Engineering, pointed out that autopilot is the development trend of future car travel. The results of the research show that if a car is hesitant to drive in a city traffic system, the speed of the car will suddenly increase and decrease, which will have a certain impact on traffic flow efficiency. Compared with human driving habits and driving techniques, automatic driving can greatly improve travel efficiency. In addition, self-driving cars will recognize and react more quickly to traffic conditions. Automated driving can reduce traffic accidents.

For a time, everyone thinks that the days of “while watching movies while driving” seem to be on the horizon. However, in reality, some key technologies for unmanned driving, especially safety issues, have not yet been completely resolved, and unmanned driving accidents that are connected together are still being repeated.

â–  Safety regulations need to be established

In recent years, thanks to the advancement of information technology and the evolution of artificial intelligence, there has been a rapid development of unmanned driving. For example, the substantial optimization of physical computing capabilities and dynamic vision technologies provides a sharp "eye" for driverless cars; while the development of artificial intelligence, digital maps, and wireless communication technologies has given cars a flexible "brain." Can make accurate driving decisions.

Despite this, many people still worry about the safety of unmanned vehicles. If there is a problem with the hardware and software of the driverless vehicle, especially in a busy urban road system, the consequences can be disastrous. Unmanned driving also inevitably triggers the so-called controversy over the "electric train problem." The skeptics argue that driverless cars cannot submit satisfactory answers when they must choose to bump into passersby or sacrifice passengers in the car. Because drone is calculated by the machine rather than the moral and emotional person. This makes many people deeply disturbed.

Industry insiders pointed out that: "At present, unmanned driving technology has matured, but there are many aspects such as information security problems, road safety problems, public awareness, technical level is not perfect, and even the problems in the moral field. Factors, all will become the threshold for the unmanned vehicle to cross the road of popularization. Of course, just as the popularization of the car, this is only a matter of time. The current need for the industry is the government departments to issue the appropriate policies and standards to regulate as soon as possible."

It is true that public confidence in driverlessness will not be established overnight. In addition to the need to overcome the challenges of unmanned technology, it is also important to speed up the process of drones by examining the self-management capabilities of vehicles, formulating standards for autonomous self-driving standards and safety assessment standards, and establishing laws and insurance claims for the handling of traffic accidents. Initiatives. Only through technical inspection, approval, and legislation to ensure driverless safety can the public “abandon the steering wheel” and enjoy the convenience and happiness of driverless driving.

At present, our government is also working on issues related to driverless safety. On June 7th this year, the first national-level smart network-linked automobile (Shanghai) pilot demonstration zone was launched. The car companies have the opportunity to test under 100 simulated complex road conditions. On June 13, the website of the Ministry of Industry and Information Technology formally released "Notice on Soliciting the Guidelines for the Construction of the National Internet of Vehicles Industry Standard System (Intelligent Gridlink Automotive) (2017) (Consultation Draft)", which initially established 95 items Drive standards and seek public comments for a period of one month.

â–  High standards and strict requirements to ensure safety

Think about this scenario: In a heavy rain or heavy fog, a car speeds up at a speed of 130km/h on the highway. As soon as the road conditions change, human drivers will be in a hurry with the assistance of driver assistance systems such as the Electronic Stability System (ESP), Lane Departure Warning (LDW) and Automatic Emergency Brake (AEB). Today, driverless cars use more than 100 processors and sensors to flexibly respond to extreme conditions, not only providing a more comfortable driving experience, but also ensuring the safety of vehicles. Of course, the premise is that all systems in the entire life cycle of the vehicle can ensure flawless and efficient operation - which imposes higher requirements on the detection and certification of driverless cars.

Among them, taking the automatic emergency braking system (AEB) regulatory requirements for driverless cars as an example, how to predict what will happen next depends on the changes in different parameters. For example: detection probability, vehicle speed, pedestrian speed, road friction, distance between cars and pedestrians, etc. Assuming that a car is approaching, various conditions and parameters will be calculated immediately. The system will quickly make driving decisions to ensure that the car and the pedestrian are safe.

So, how does the driverless security guarantee? It is particularly important to develop a complete set of assessment methods based on real-world scenarios for the system development process, legislative advice, and type approval and product safety verification.

This assessment method provides a comprehensive analysis of the scene, environment, system configuration, and driver characteristics. It identifies some important and complete test items and conducts evaluations with different test levels, including virtual, verification, and on-site testing. First, scenarios are determined based on assumptions. For example, traffic peaks, heavy fog, and overtaking in mega-cities during the evening. Then conduct a virtual test, and then conduct a track test at the test site to verify the virtual test results, and finally through the field test for scenario verification.

For example, current standards that ensure the interaction of conventional braking and steering (ECE R13/-79) lateral and horizontal assistance functions no longer comply with next-generation auxiliary system specifications. Therefore, experts from TÃœV SÃœD are collaborating with the Special Committee on Driver Assistance Systems of the German Federal Ministry of Transport and Digital Infrastructure to update relevant specifications. Experts estimate that 100 million scenarios must be tested before a fully autonomous driving function is approved.

â–  Information security is a top priority

“In the past we only considered driving safety, including active safety and passive safety. However, today’s information cannot be ignored.” On the issue of driverless safety, Yin Chengliang, deputy dean of the Automotive Engineering Research Institute of Shanghai Jiaotong University, emphasized the information security of smart cars. .

In addition to the test evaluation of driverless cars, the safety of smart car transportation systems is also crucial. Similar to other IoT devices such as smart phones and wearable devices, self-driving cars will also generate consumption data, which will also affect the way companies develop and influence government decisions. However, unlike other networked devices, if the software and hardware infrastructure of a self-driving car has problems, then the risk of data exposure is even greater.

Mobile phones and computers are hacked and may reveal privacy, but at least it will not directly endanger personal safety. However, the car may not be the same, it carries people to travel, once the hidden dangers, the consequences could be disastrous. Previously, there have been many accidents involving cars being hacked. Among them, two hackers, Charlie Miller and Chris Valasek, hacked the Chrysler vehicle system. They connect wirelessly to the car and can easily remotely control the vehicle's brakes, steering and shifting functions with just a laptop whenever and wherever they are. Although the two hackers emphasized that their "invasion behavior" is based on research, the purpose is to point out that there are vulnerabilities in the car and the relevant personnel to solve this problem, but the consequences of this will make the entire automotive industry restless. The loophole they discovered directly led Chrysler to recall more than 1.4 million cars. In addition, news of invading vehicles through entrances such as OBD, car systems, and remote control keys is also emerging.

Regarding the issue of automobile information security, executives of automobile companies also admitted that the current auto industry is lagging behind in this regard. However, many auto companies have begun to pay attention to information security issues and started to develop conventional safety measures. With the popularity of unmanned vehicles with higher levels of vehicle intelligence in the future, more and more new features will be added. It seems that the network security loopholes in cars may become more and more, and solutions will increase. The more troublesome. Even so, it still needs continuous efforts. After all, it is a matter of security, and it cannot be over-invested.