An Integrated Strategy for Autonomous Exploration of Spatial Processes in Unknown Environments.

Area of Science:

• Robotics and Artificial Intelligence
• Spatial Data Analysis
• Environmental Monitoring

Background:

• Robotic exploration is crucial for spatial processes like radioactivity or temperature.
• Existing methods often assume known environments, limiting real-world applicability, especially in disaster scenarios.
• Integrating map and process exploration presents conflicting goals.

Purpose of the Study:

• To develop a novel integrated strategy for robots to explore spatial processes in unknown environments.
• To fuse spatial process modeling with robot mapping and localization.
• To address the conflict between map exploration and process exploration goals.

Main Methods:

• Utilized Gaussian Processes (GP) to model the spatial process of interest.
• Employed process entropy to guide exploration.
• Integrated registration algorithms for robot mapping and localization.
• Used frontier-based exploration for environmental mapping.
• Developed a trade-off strategy to balance process and map exploration.

Main Results:

• Extensive evaluations in simulated environments demonstrated superior performance compared to baseline strategies.
• Experimental verification with a mobile robot in a labyrinth environment confirmed the strategy's effectiveness.
• The integrated strategy outperformed both frontier-based and GP entropy-driven exploration methods.

Conclusions:

• The proposed integrated strategy effectively enables robots to explore spatial processes in unknown environments.
• The trade-off mechanism successfully balances competing exploration objectives.
• This approach offers a significant advancement for robotic applications in complex, unmapped terrains.