Chapter 9 Computing Systems
Autonomous vehicles (AVs) represent the convergence of several advanced technologies, combining robotics, artificial intelligence, control theory, and embedded systems engineering into a cohesive, real-time decision-making platform. The computational demands of AVs far exceed those of conventional vehicles, requiring the seamless orchestration of heterogeneous software and hardware systems to operate safely and reliably in dynamic environments.
Unlike human drivers, who continuously perceive, reason, and act in real-time, AVs must emulate this cognitive loop using a structured computational pipeline. This pipeline includes distinct yet interdependent stages: sensor data acquisition, perception and scene understanding, localization within a mapped environment, motion planning, and vehicular control. Each of these stages imposes strict latency and reliability requirements on the underlying computing infrastructure. Moreover, the pipeline must be capable of operating continuously, in parallel, and in real-world conditions marked by uncertainty and complexity.
Additional Readings
Liu, L., Lu, S., Zhong, R., Wu, B., Yao, Y., Zhang, Q., & Shi, W. (2020). Computing systems for autonomous driving: State of the art and challenges. IEEE Internet of Things Journal, 8(8), 6469-6486.
Liu, L., Liu, S., & Shi, W. (2021). 4C: A computation, communication, and control co-design framework for CAVs. IEEE Wireless Communications, 28(4), 42-48.
Wu, T., Wu, B., Wang, S., Liu, L., Liu, S., Bao, Y., & Shi, W. (2021). Oops! it’s too late. your autonomous driving system needs a faster middleware. IEEE Robotics and Automation Letters, 6(4), 7301-7308.