Risk Management System Based on Uncertainty Estimation by Multi-Robot
Daichi Kato, Kosuke Sekiyama, Toshio Fukuda · Journal of Robotics and Mechatronics · 2010
The risk management we propose uses multi-robot patrols to maintain security, which we treat as equivalent to minimizing observational uncertainty of a place - we call this place checkpoint i (i = 1,2, . . . ,n). We therefore formulate the uncertainty by entropy in information theory. Robots patrol and observe the checkpoint’s condition and update the patrol schedule based on the estimated uncertainty of checkpoints in real time. To relieve uncertainty, we propose Earliest Deadline First scheduling with adaptive Risk Estimation (EDFRE), then compare EDFRE with simple EDF scheduling and evaluate EDFRE adaptability to changes under different initial conditions. Results demonstrated EDFRE’s effectiveness in dynamic situations.