Autonomous driving strategies for urban and highway environments

点击翻译按钮获取中文摘要

The relentless expansion of sensing, computing and communication capability being realized due to ever more capable integrated circuits has revolutionized industry after industry. The automotive sector is poised to be one of the next major area of revolution, as vehicles learn to drive themselves, increasing safety, efficiency and passenger comfort. This research project seeks to help bring about this revolution in autonomous driving by maximally exploiting the unrealized potential benefits vehicle to vehicle and vehicle to roadside communication as well as vision, laser ranging and radar sensors in an economical and practical manner. The work will focus on highway vehicle platooning, emergency braking scenarios, and static and dynamic object detection, enabling most of the key human driver capabilities currently not available to the general public. Highway platooning and emergency braking will be advanced by exploiting low-latency communication to create multi-lane configurations and mixed-equipage solutions. Customized, dedicated hardware solutions including vision sensors and FPGAs will enable real time object tracking, online sign and stoplight detection, and pedestrian collision avoidance. To enable rapid validation of new algorithms and hardware components, a unique 1/6th scale autonomous driving vehicle fleet will be constructed, complete with low-latency communication, sufficient onboard computation power, and all the necessary sensors. These electric vehicles will be capable of operating at 50 kph and will include steering, braking and suspension components to more accurate mimic true on-road driving conditions.