Large variations in the execution times of algorithms characterize many cyber-physical systems. Variations arise in visual object-tracking tasks, when execution times depend on the contents of the current field of view of the camera. The authors studied this scenario in a small unmanned aerial vehicle (UAV) system with a camera that must detect objects in a variety of conditions. Given resource, weight, and size constraints, such cyber-physical systems do not have the resources to satisfy the hard-real-time requirements of safe flight while processing variable workloads at high quality and resolution levels. Tradeoffs must be made. Specifically, the usefulness of tracking an increasing number of objects may saturate when mission software can no longer perform the required processing on each object.
This material was presented at the Third IEEE/ACM International Conference on Cyber-Physical Systems in Beijing, China. The authors evaluate a new approach called ZS-QRAM (zero-slack quality-of-service-based resource allocation model) that maximizes the UAV system utility by explicitly accounting for the diminishing returns on tracking an increasing number of objects. The conference paper includes a detailed evaluation of the approach on a UAV system to clearly demonstrate its benefits.