A Bayesian Approach to Risk-Based Autonomy for a Robotic System Executing a Sequence of Independent Tasks
Enabling higher levels of autonomy requires an increased ability to identify and handle internal faults and unforeseen changes in the environment. This work presents an approach to improve this ability for a robotic system that is executing a series of independent tasks, such as inspection, sampling, or intervention, at different locations. A dynamic decision network (DDN) is used to infer the presence of internal faults and the state of the environment based on the available measurements. This knowledge is used to evaluate the risk of executing the current task, which is used to evaluate whether the task should be executed or skipped and whether maintenance actions are needed. Evaluating past states given new information is used to identify skipped tasks that should be revisited. The proposed approach is implemented for a drone tasked with contact-based ultrasound inspection of an industrial facility. The drone is able to successfully distinguish between different internal faults and adverse environmental states and act accordingly. The system makes risk-informed decisions based on uncertain knowledge, enabling it to minimize the time usage while minimizing the potential of harming the drone and maximizing mission completion.<br>