教授
Loughborough University, UK
暂无信息
Fellow of IEEE EPSRC Established Career Fellow Chair in Autonomous Vehicles
Abstract: There is a huge aspiration about intelligent vehicle technologies including ADAS and driverless cars. Despite of all the significant progress so far, there are still many fundamental challenges that must be addressed before the potential of intelligent vehicles could be fully realized. Among others, verification of the safety of technologies, the interaction of many sophistic functions involved in intelligent vehicles, and the complexity of driving environments and human factors are on the top list. This talk argues, to address these challenges, it is more important to take a system level approach, rather than only at the algorithm level. A Goal-Oriented Control System (GOCS) is proposed to facilitate the analysis and design of intelligent vehicles at a system level where feedback plays a central role. Examples are given to illustrate how it helps to understand the interaction between different core functions (e.g. the interaction between optimization algorithms (e.g. for path planning and machine learning) and physical dynamics). Another feature of this framework is to promote goal-oriented approaches, moving away from scenario-based description approaches. It is the control system to decide, given all the available information for a specific scenario, what is the best course of action to achieve the driving goal while respecting to all the constraints such as safety, and the rules of the road. The talk concludes with discussion on future research in this area.
