SUMMARY

The research objective of this work is to calculate, measure, and demonstrate the performance increase in terms of energy efficiency and operator productivity enabled by a novel shared control algorithm applied to hydraulic excavation. The testbed represents a class of human-controlled, multiple degree-of-freedom manipulators and thereby ensures the research is germane to the robotics field in general. Presently, excavator operators manually control end effector motion. Decreases to a task's time or energy cost are historically achieved by designing better systems or components or through operator training. In the proposed approach, the human and machine share control of the end effector to improve machine performance. Here, the operator is continually in the loop, meaning an operator input immediately elicits changes in actuator motion. This is in contrast to impractical autonomous machines for which pre-programmed trajectories remove the operator from the control loop. The proposed shared control scheme is exemplified by a controller having authority to perturb the original command, thereby sharing control with the operator. The command perturbation is a function of estimates of the intended task and the pseudo-optimal trajectory for completing the task. The proposal presents justifications of three hypotheses: that expert operators currently fail to run manually controlled machines at peak performance; that a shared controller can perturb an operator command to cause the end effector to follow a better trajectory; and that this control paradigm is stable and acceptable to the operator.