Motion planning of cleaning robot based on 3D vision

Area of Science:

• Robotics and Automation
• Computer Vision
• Industrial Engineering

Background:

• Traditional industrial cleaning faces challenges like water waste, blind spots, and poor equipment compatibility.
• Existing methods often lack adaptability to diverse materials and stain types.
• Automated cleaning solutions are needed to enhance efficiency and quality in manufacturing.

Purpose of the Study:

• To develop a 3D vision-based cleaning robot for industrial applications, specifically in rail transit equipment manufacturing.
• To address limitations of current cleaning methods by integrating intelligent control and advanced recognition algorithms.
• To improve cleaning efficiency, quality, and resource management (e.g., water usage).

Main Methods:

• A three-dimensional vision system integrated with a cleaning robot.
• Proportional-Integral-Derivative (PID) control for precise adjustment of cleaning pressure and water temperature.
• Improved CIoU-YOLOv7 object recognition and template matching for workpiece and stain identification.
• Genetic algorithm for cleaning path optimization.

Main Results:

• Achieved a workpiece recognition accuracy of 62.34%, a 5-10% improvement over baseline algorithms.
• Optimized cleaning path reduced total path length by 64.7% and increased efficiency by 10.5% compared to basic genetic algorithms.
• Reduced the number of generations for optimal solutions by 95%, indicating enhanced intelligence and efficiency.

Conclusions:

• The proposed 3D vision-based cleaning robot effectively addresses limitations in industrial cleaning, including water waste and inefficiency.
• The integration of advanced algorithms enhances workpiece recognition, positioning, and cleaning path optimization.
• The system offers significant improvements in cleaning efficiency, quality, and resource conservation for manufacturing applications.