A fuzzy-behavior-based approach for controlling mining excavator bucket/rock interactions

Successful control of an autonomous mining excavator requires sensor feedback from the excavation tool to control unpredictable tool/rock interactions. Conventional control approaches cannot deal effectively with uncertainties in the dynamic and unstructured mining environment. This paper presents a new intelligent control approach that utilizes fuzzy behaviors with fuzzy strategies to control bucket motions for autonomous excavation tasks. In this approach, the control system uses feedback from a bucket mounted force/torque sensor to control excavation motions by sensing and reacting to interactions with rock particles during excavation. A behaviour coordination configuration is developed to improve the controllers effectiveness in reaching its overall excavation goals. It uses fuzzy logic rules to adjust control system output variables and performs goal-driven behaviour selection. This technique is a first step towards realistic control of complex machine/environment interactions in difficult and unpredictable environments, where traditional control approaches are either impossible or computationally excessive. The discussion here focuses on: (a) development of a new control architecture; (b) the individual components of the architecture; and (c) experiments with a simulated excavator using a PUMA 560 robotic arm