Heuristic‐Based Architecting for Autonomous Vehicle Systems

Abstract

AbstractHumans have always had a need to improve travel, whether that be for ourselves or for the goods we need. Technological advancements have all fundamentally performed the task of moving people and goods with the objective to continuously improve convenience and be more efficient. There is another pending revolution in human mobility, one that is leading towards autonomous transportation where the human driver is replaced, and the driving of the vehicle is performed autonomously by machine. With continued research and technology, automotive original equipment manufacturers (OEMs) have started to implement autonomous enabled features to transform today's vehicles traveling our roads into driverless vehicles. This paper evaluates the fundamental problem of architecting a system responsible of transporting occupants and cargo from point A to point B via autonomous vehicles (AVs) and compares it to current architecture being proposed/produced by OEMs. The work begins by studying the user needs of an AV and mapping those needs to required system functions. Then, guiding heuristics and lessons learned are used to critique the current architecture. The results show that the current OEM approach is to simply create smarter vehicles within a non‐intelligent infrastructure, and the authors, when guided by system architecting heuristics, suggest that development of an elegant, robust, long‐term AV system will require a significant increase in scope to a higher system level than the OEMs appear to be addressing at the present time.