Bio-Inspired Vision and Gesture-Based Robot-Robot Interaction for Human-Cooperative Package Delivery

Area of Science:

• Robotics
• Computer Vision
• Human-Robot Interaction

Background:

• Cooperative robotics often relies on complex network communication.
• Constrained environments pose challenges for inter-robot coordination.
• Human-in-the-loop systems enhance robotic task execution.

Purpose of the Study:

• To develop a novel bio-inspired framework for network-free robot-robot interaction.
• To enable cooperative package delivery between two robots with human guidance.
• To eliminate the need for network-based communication in constrained environments.

Main Methods:

• Utilized object detection (custom YOLOv4) and depth perception for robot-robot communication.
• Developed a shape identification method based on calculating the area of detected polygonal routes.
• Analyzed gestures for accuracy in conveying direction, distance, and target coordinates.

Main Results:

• Achieved 93.3% accuracy in simulations and 90% in physical setups for shape identification.
• Demonstrated an average positional Root Mean Square Error (RMSE) of 0.349 in simulation and 0.461 in physical experiments.
• Successfully implemented a cooperative package delivery system without network reliance.

Conclusions:

• The proposed bio-inspired framework effectively enables network-free cooperative robotics.
• Visual cue inference and gesture recognition are viable methods for inter-robot communication.
• The system shows promise for real-world applications in constrained environments.