VP-SOM: View-Planning Method for Indoor Active Sparse Object Mapping Based on Information Abundance and Observation...

Area of Science:

• Robotics
• Artificial Intelligence
• Computer Vision

Background:

• Active mapping is crucial for mobile robot autonomy in indoor environments.
• Current methods often overlook object-level details, hindering human-robot interaction.
• View planning significantly impacts map quality and exploration efficiency.

Purpose of the Study:

• To propose a novel view-planning method for indoor active Sparse Object Mapping (VP-SOM).
• To incorporate object cluster properties for enhanced human-robot environments.
• To balance exploration efficiency and mapping accuracy using categorized views.

Main Methods:

• Categorized views into global and local based on object clusters.
• Developed a new view-evaluation function using object information abundance and observation continuity.
• Introduced an object surface occupancy probability map to calculate sparse object model uncertainty.

Main Results:

• VP-SOM demonstrated more accurate and efficient indoor environment exploration.
• The method successfully built robust object maps.
• Experimental results validated the effectiveness of the proposed view-planning strategy.

Conclusions:

• The VP-SOM method advances active mapping by integrating object-centric information.
• This approach improves the accuracy, efficiency, and robustness of indoor object mapping.
• The findings are significant for developing more interactive and capable mobile robots.