We still have a long way to go to achieve a road full of autonomous vehicles. Although the current sensor capabilities are very helpful, they are relatively rudimentary when it comes to autonomous vehicles. In other words, in the past few years, the automotive industry has made tremendous progress. If you launched a brand new car five or six years ago, it may contain about 60 to 100 sensors. Today, this number is close to 200 or more. As vehicles continue to become more intelligent, the evolution and complexity of sensors will also change, and will continue to grow as new features become available. In order to promote the development of autonomous vehicles, technology suppliers and automakers are considering various challenges, such as sensor degradation, cooperation with industry standards, and maintenance/maintenance of software cyber security defenses during the vehicle's life cycle.
Competition in the automotive industry
One of the biggest challenges facing OEMs is to keep up with the rapid pace of sensor and data development. The sensor needs to provide the necessary data confidence for the automotive system to meet the design requirements. Driving on dense metropolitan roads filled with a large number of drivers, pedestrians, bicycles and motorcycles requires vehicles to make instant decisions, stop or turn to avoid hitting pedestrians or other vehicles. Therefore, we are beginning to see sensor fusion play a role, in the time of ADAS response, by consuming and interpreting various data inputs, making decisions that human drivers may not be able to make.
In the initial stage of sensor manufacturing, car manufacturers must also comply with automotive functional safety standards to prevent chip or software failure. ISO 26262 is a vehicle regulatory standard that specifies the development process that OEMs and suppliers must follow to achieve functional safety standards. By complying with ISO 26262, OEMs and suppliers provide assurance that their equipment will perform within the expected time. The challenge of sensor degradation Another aspect of the ever-changing sensor field is sensor degradation. Sensor degradation is a natural part of self-driving cars, especially considering that today's vehicles usually have a lifespan of 10 to 15 years. The main causes of degradation include general wear and tear of the sensor, harsh working environment, and degradation of other electronic system components.
Automakers and technology suppliers need to consider whether sensors such as lidar, cameras, and ultrasonics can achieve the same or even higher performance levels as new cars during the entire expected life of the vehicle. They must also answer what happens if the sensor starts to fail (ie, how to alert the driver, establish safety features, etc.). To combat degradation, OEMs need to model and design semiconductors and other components in vehicles to create predicted failure rates and alternatives in many different environments.
Network security troubles
There are several different factors that need to be considered when ensuring the safety of sensors in autonomous vehicles. Hacking into self-driving cars is always a worrying issue and should be addressed, but another less obvious security factor is that the attacker influences the machine learning technology embedded in the vehicle to make it react in a malicious way. For example, a British study introduced a video billboard that was tampered with to display a stop sign for as little as a fraction of a second. Self-driving vehicles will sense the stop sign and stop because they can receive the image and react to it in the same way that they sense a stop sign on the road. However, the human driver will not be able to react to the image. If the autonomous vehicle in front stops suddenly without warning or obvious reason, it will be potentially dangerous. These "phantom objects" may harm drivers and pedestrians. This is just another challenge facing automakers in ensuring the safety of autonomous vehicles.
In order to strengthen the prevention of sensor vulnerabilities, NHTSA recommends that the automotive industry follow the cyber security framework documented by the National Institute of Standards and Technology (NIST). The framework proposes a layered approach to cyber security around these key functions: identification, protection, detection, response, and recovery. The system designer must ensure the operation of the system while striking the right balance in sharing alerts with drivers. With ADAS in mind, it is designed to warn drivers of impending dangers on the road and potential system failures. Under normal circumstances, system errors caused by security vulnerabilities will trigger a recall, which requires going to the dealer, which may be an expensive repair cost for the car factory. A better method is to build on security to prevent malicious behavior from causing errors in the first place.
As the complexity and capabilities of these devices increase, different ways of network security attacks will increase accordingly. Designers and car manufacturers will need to update their defenses to manage these types of unexpected reactions to prevent these non-traditional cyber attacks. Part of this defense should be a strong software lifecycle management plan that allows organizations to apply the lessons learned from their experience. This concept also applies to the physical properties of the sensor itself to prevent degradation.
As self-driving cars become more and more common, there will be more regulations and various consumer habits that will help shape the future of the market. As we continue to see self-driving cars enter the mainstream, keeping pace with advances in the field of sensor fusion is challenging but necessary.