An autonomous excavator system for material loading tasks

Area of Science:

• Robotics and Automation
• Artificial Intelligence in Construction

Background:

• Excavators are crucial for material handling in unstructured mining and construction environments.
• Real-world operation poses risks due to extreme conditions, leading to injuries and fatalities.

Purpose of the Study:

• To develop and evaluate an autonomous excavator system (AES) for material loading tasks.
• To enhance safety and efficiency in hazardous excavation operations.

Main Methods:

• A multimodal perception system fusing LiDAR, cameras, and advanced image processing.
• Hierarchical task and motion planning integrating learning-based and optimization methods.
• Integration of perception, planning, and control modules for robust operation.

Main Results:

• AES demonstrated efficient and autonomous handling across diverse scenarios like material loading, waste handling, and trenching.
• The system achieved 24 hours of continuous operation per intervention.
• Material handling rates comparable to experienced human operators were observed.

Conclusions:

• The developed autonomous excavator system offers a robust solution for material handling in challenging environments.
• AES significantly improves operational efficiency and safety in mining and construction.
• The system's performance is comparable to human operators, paving the way for wider adoption.