ABSTRACT:In this paper, we develop an algorithm for avoidance of earth obstacles using model-reference impedance control for autonomous excavator. There have been many research in the past regarding autonomous excavation, which were only concerned about position control, force control and path generation. However in order to apply the control system or algorithm to the actual autonomous excavation, it is essential that prediction and avoidance of potential obstacle must be carried out. When earth obstacles such as rocks and tree roots are detected by force/torque sensor, accidents can be easily prevented by halting the excavation task using a simple fail-safe algorithm. However, it would be more desirable to control the unpredictable bucket/earth obstacle interactions intelligently on-line while continuing the excavation task instead of stopping. Impedance control is a control scheme, which is used for defining dynamic relationship between end-effector and environment when the end-effector is constrained in the environment. In this case end-effector would be the bucket tip and environment would be the earth obstacle. In this paper a novel algorithm was developed which realize the obstacle avoidance of autonomous excavator that can regulate bucket force in real time.For the experimental use, hydraulic manipulator which resembles commercial excavator was designed. Two kinds of sensor, potentiometers and 6 axis force/torque sensor, were used to provide position and bucket tip force feedback information. Experimental results indicate that bucket has avoided earth obstacle smoothly without large force and all desired task is completed.