Fast UAV Object-Searching in Large-Scale and Complex Environments.

Area of Science:

• Robotics
• Computer Vision
• Artificial Intelligence

Background:

• Autonomous object-searching is vital for unmanned aerial vehicles (UAVs).
• Existing methods often explore inefficiently, either focusing solely on unknown areas or repeating searches.
• There is a need for improved strategies in large-scale, complex environments.

Purpose of the Study:

• To introduce an effective object-searching strategy for UAVs in large-scale and complex environments.
• To enable fast, secure, and efficient object searches.
• To enhance autonomous exploration capabilities.

Main Methods:

• A Kalman filter-based YOLO algorithm for robust object position estimation in cluttered and occluded scenes.
• A mode-based method for computationally efficient viewpoint generation.
• A hierarchical searching strategy integrating global path planning and local trajectory optimization using frontier data.

Main Results:

• The proposed method significantly reduces both searching and computation times compared to existing techniques.
• Experimental results across six diverse environments demonstrate superior performance.
• The strategy proved effective in various real-world scenarios.

Conclusions:

• The novel object-searching strategy enhances UAV efficiency in complex environments.
• The method offers a robust and computationally efficient solution for autonomous exploration.
• This approach represents a significant advancement in autonomous object-searching for UAVs.