Collision Control and Avoidance for Electric Mining Shovels

This report details the development and demonstration of a collision control technology for P&H-class mining shovels that minimizes the frequency and severity of dipper collisions between the bucket, crawler tracks and floor. The work was funded by ACARP as project C15038.

The collision control technology uses a Model Predictive Control (MPC) framework whereby control actions for preventing collisions, are chosen based on the predicted future behaviour of the shovel. The system predicts future locations of the dipper over a receding time horizon using a set of mathematical models that are driven by the operator’s joystick commands. The prediction models incorporate the shovel’s electro-mechanical dynamics for hoist, crowd and swing motions.

Predicted future bucket positions are tested for collisions using computational geometry algorithms on VRML models of the bucket, crawler tracks and floor.

If a collision is predicted, the collision prevention algorithms calculate the minimum change to the operator’s command that will avoid the collision by a specified clearance distance. The altered joystick references are then passed to the P&H Centurion control system over the shovel’s PROFIBUS control bus.

The technology has been successfully demonstrated on the CRCMining P&H 2100BLE research shovel at Bracalba Quarry in Queensland and on a P&H 4100A at Caballo Mine in Wyoming USA.

The technology is currently being commercialized by P&H Mining Equipment.