Vehicle Autonomy to Ecosystem Intelligence: A Systematic Review of Dynamic Vision Architectures in Surface Mining...

Area of Science:

• Mining Engineering
• Robotics
• Artificial Intelligence

Background:

• Autonomous Haulage Systems (AHS) in surface mining currently use vehicle-centric perception.
• Limitations exist in handling dynamic mining environments (dust, terrain, traffic).
• Egocentric perception models struggle in complex mining ecosystems.

Purpose of the Study:

• To systematically review dynamic vision systems for AHS in surface mining.
• To analyze the shift from autonomy to interconnected, ecosystem-aware intelligence.
• To propose a framework for Ecosystem-Centric Dynamic Vision (ECDV).

Main Methods:

• Systematic literature review synthesizing research from mining automation, robotics, intelligent transportation, and multi-agent perception.
• Critical analysis of sensing technologies, perception algorithms, sensor fusion, and environmental robustness.
• Development of a conceptual framework for ECDV.

Main Results:

• Identified limitations of current egocentric perception in AHS.
• Proposed ECDV framework integrating fleet communication, dispatch, digital twins, and environmental sensing.
• Outlined a multi-layer architecture for cooperative perception and predictive hazard modeling.

Conclusions:

• Transitioning AHS from vehicle autonomy to ecosystem intelligence is crucial.
• ECDV enhances perception through integration with mine-wide data sources.
• Future research should focus on cooperative perception, adaptive sensor fusion, and digital-twin-integrated safety systems.