A Multirobot Person Search System for Finding Multiple Dynamic Users in Human-Centered Environments.

Area of Science:

• Robotics
• Artificial Intelligence
• Optimization Algorithms

Background:

• Multirobot coordination is crucial for applications like search and rescue and daily living assistance.
• Current methods often lack robot coordination in team planning or fail to integrate user location probabilities, leading to inefficient searches.

Purpose of the Study:

• To present a novel multirobot person search system for efficient multi-user detection within a deadline.
• To enhance search efficiency by simultaneously considering robot actions and dynamic user location probabilities.

Main Methods:

• A two-stage optimization problem is formulated to maximize the expected number of users found.
• Stage 1 uses a min-flow graph and Bellman-Ford algorithm for action selection.
• Stage 2 employs a fuzzy clustering method for action allocation, modeling the environment as a search region network.

Main Results:

• The proposed system outperforms existing approaches in finding more target users before the deadline.
• Experiments demonstrated effectiveness across varying environment sizes, search durations, and numbers of users.
• The simultaneous consideration of robot actions and user probabilities significantly improves search outcomes.

Conclusions:

• The novel multirobot search system offers a significant advancement in coordinated search operations.
• This approach provides a more effective strategy for locating multiple dynamic users in complex environments.
• The developed optimization techniques offer a robust solution for real-world robotic search applications.