Cooperative Active Learning-Based Dual Control for Exploration and Exploitation in Autonomous Search

Area of Science:

• Robotics and Autonomous Systems
• Artificial Intelligence
• Signal Processing

Background:

• Autonomous search in unknown environments with unknown sources presents significant computational challenges.
• Existing methods often rely on computationally intensive nonlinear Bayesian estimation and complex decision-making.
• There is a need for efficient algorithms that balance performance with computational feasibility.

Purpose of the Study:

• To develop a computationally efficient algorithm for autonomous search and source estimation in unknown environments.
• To introduce a cooperative active-learning-based dual control for exploration and exploitation (COAL-DCEE) framework.
• To reduce the computational burden associated with traditional Bayesian estimation approaches.

Main Methods:

• Deployment of multiple cooperative estimators for environment learning and source estimation.
• Implementation of a dual control strategy for simultaneous exploration and exploitation.
• Analysis of algorithm convergence and performance concerning sensor noise and turbulence.

Main Results:

• The proposed COAL-DCEE algorithm significantly reduces computational load compared to information-theoretic approaches.
• Multiple cooperative estimators enhance search performance and robustness against noisy measurements.
• Convergence and performance guarantees are established for the COAL-DCEE framework.

Conclusions:

• COAL-DCEE offers a computationally efficient solution for autonomous search and source estimation.
• The framework provides convergence guarantees and comparable search performance with reduced computational power.
• This approach is effective in unknown environments with sensor noise and turbulence disturbances.