Towards computational awareness in autonomous robots: an empirical study of computational kernels

Abstract The potential impact of autonomous robots on everyday life is evident in emerging applications such as precision agriculture, search and rescue, infrastructure inspection. However, necessitate operation unknown unstructured environments with a broad sophisticated set objectives, all under strict computation power limitations. We therefore argue that the computational kernels enabling robotic autonomy must be scheduled optimized to guarantee timely correct behavior, while allowing for reconfiguration scheduling parameters at runtime. In this paper, we consider necessary first step towards goal awareness robots: an empirical study base from resource management perspective. Specifically, conduct data-driven timing, power, memory performance localization mapping, path planning, task allocation, depth estimation, optical flow, across three embedded computing platforms. profile analyze these provide insight into dynamic computation-aware robots. Notably, our results show there correlation kernel robot’s operational environment, justifying notion why work crucial goal.