Sensor Fusion-Based Cooperative Trail Following for Autonomous Multi-Robot System

Area of Science:

• Robotics
• Computer Vision
• Artificial Intelligence

Background:

• Autonomous trail following in natural environments is a complex challenge for robotic systems.
• Current deep learning methods primarily focus on single-robot vision-based trail following.
• Real-world robotic applications, like search and rescue, often involve multi-robot cooperation.

Purpose of the Study:

• To develop a cooperative trail-following method for multi-robot systems using sensor fusion.
• To enhance the robustness and accuracy of autonomous navigation in wild environments.
• To enable robots to share and fuse sensor data for collective decision-making.

Main Methods:

• A sensor fusion-based cooperative trail-following approach is proposed.
• Robots exchange vision data and fuse sensor information from different altitudes.
• A "threshold" mechanism is employed to manage the sensor data fusion process, optimizing for quality of service.

Main Results:

• The cooperative method significantly improves recognition accuracy compared to single-robot systems.
• Experiments demonstrate enhanced robustness in real-world datasets.
• Collective-level fusion of vision data features leads to better trail-following performance.

Conclusions:

• Cooperative sensor fusion enables multi-robot systems to achieve more robust and accurate autonomous trail following.
• The proposed method offers a practical solution for complex robotic navigation tasks in unstructured environments.
• Integrating collective intelligence through sensor fusion is key to advancing multi-robot system capabilities.