Implementation of two-roller scheduling path planning under road construction scenarios

Area of Science:

• Robotics and Automation
• Civil Engineering
• Construction Management

Background:

• Previous research focused on single unmanned roller path planning, neglecting multi-roller coordination and faulty machine scheduling.
• Scheduling issues with faulty rollers (e.g., low oil pressure) are common in collaborative pavement construction but lack comprehensive strategies.

Purpose of the Study:

• To develop a framework and scheduling strategy for idle and faulty robots in a multi-roller pavement construction system.
• To address the challenge of scheduling faulty rollers by defining a methodology for determining robot start and end positions.

Main Methods:

• Introduced a framework for pavement construction with one paver and six rollers.
• Developed a methodology using Bezier curves and idle region points to determine scheduling positions for faulty and idle robots.
• Proposed a novel cost function balancing path length and safety distance, considering scene and model constraints.
• Implemented an interleaved scheduling scheme to handle narrow passages and improve solution feasibility.
• Utilized discrete sampling for efficient algorithm execution.

Main Results:

• The proposed path planning algorithm is more effective than alternatives for specific application scenarios.
• The interleaved scheduling framework significantly enhances the quality of generated paths.
• Successfully scheduled faulty and idle rollers for over 100m road length (10.5m width) without impacting other rollers.
• Achieved a path search time of 24s, planned path length of 178.1m, and maintained a safety distance of approximately 1.6m.

Conclusions:

• The developed scheduling strategy effectively addresses the problem of faulty and idle roller management in multi-robot pavement construction.
• The methodology provides a practical and efficient solution for complex scheduling scenarios, improving overall construction quality and performance.