Page - (000606) - in Autonomes Fahren - Technische, rechtliche und gesellschaftliche Aspekte

Product Liability Issues in the U.S. and Associated Risk Management586 ‡ Mechanical or physical defects in various systems of the vehicles or their safety equipment, such as the use of materials that are not strong or thick enough,14 or an excessively high center of gravity subjecting the vehicle to rollovers. ‡ Defects in electrical components or systems other than sensors or control systems for autonomous driving, such as the use of wrong kind of components, problems in the performance of the components, or the lack of durability of the components. ‡ Software15 defects relating to systems other than sensors or control systems for autonomous driving, including information security vulnerabilities. These defects will occur in both conventional and autonomous vehicles and thus existing law and litigation methods would apply to determine a manufacturer’s liability. Nonetheless, AVs may experience defects that conventional vehicles do not. Again, they may be mechanical, electronic, or software. ‡ Mechanical or physical defects in the control systems for autonomous mode or the sensors used by the autonomous systems. A simple example would be weak mount- ings for LIDAR sensors which, if they failed, might cause the AV to lose its sensor data suddenly and crash. ‡ Defects in electrical components for sensors or control systems for autonomous driving. ‡ Software defects in the sensors or control systems used for autonomous mode. The most interesting and perhaps most concerning potential defects are those in the soft- ware used for autonomous driving. Some examples include: ‡ Designs that depend on inadequate data from sensors, including insufficient amount, inaccuracy, deficient precision, or inadequate speed of data input. ‡ Inaccurate pattern recognition, such as the AV failing to be able to recognize a pedestrian in the road or other upcoming obstacles or hazards. ‡ Designs that fail to perform safe ordinary maneuvers such as turns, lane-keeping, distance-keeping, and merging. ‡ Other problems with autonomous behavior, such as unpredictable changes in speed or direction. ‡ Deficient collision avoidance algorithms. ‡ Information security vulnerabilities. ‡ Defects arising from inadequate human-computer coordination. For instance, if an AV switches between autonomous and manual mode, the AV must alert the driver before switching to manual mode and transition to human control safely. 14 For example, the author was involved in one case in which the plaintiff alleged that the metal in a car’s tie rod was not strong enough, the metal fatigue experienced by ordinary wear of the car weakened the tie rod, and an accident occurred because metal fatigue caused the tie rod to break. 15 The “software” involved may be in the form of code built into hardware or firmware.