Font Size: 

Resource-constrained continuous surveillance tasks represent a promising domain for autonomous robotic systems. However, solutions to these problems must operate within a high-dimensional space and have an accurate model of the underlying target distribution to perform well. We present an integrated human, autonomous planning system that iteratively explores the space of surveillance solutions based on limited human interactions in an attempt to maximize both quantitative task-performance and qualitative operator satisfaction measures. This paper includes a description of our approach, implementation, and a demonstration that illustrates representative planning.

autonomy, intelligence, surveillance, human-robot interaction