This paper proposes an approach to human-guided teleoperation of a robot arm manipulator based on the configuration space (C-space) rather than on the common work space (W-space) control. Instead of directly confronting the problem of collision analysis, which is known to be extremely challenging for the human spatial reasoning, the task is offered to the operator in C-space where one can concentrate on global navigation, leaving collision analysis to the computer. Thus reduced task becomes a maze-searching problem in which humans are known to be very good. Designing such a system takes, first, calculation of the C-space, and second, an adequate user interface.
While this approach can be immediately useful even in its two-dimensional version described, in order to become a truly universal tool it needs to be extended to the three-dimensional case and to more degrees of freedom. The advantage for the operator of dealing with a point rather than a complex jointed kinematic structure is obvious. The challenge is to produce an adequate user interface (specifically, develop ways of visualizing and guiding a point in a higher-dimensional space) and to do C-space calculation and collision analysis fast enough to keep the operator active at the control station. One possibility here is to help the operator handle the environment with incomplete, rather than complete, information; this would mean a significant reduction in the C-space computation costs. Success in this area will also mean applicability of the approach to a dynamic environment with moving obstacles. Computer algorithms for motion planning with incomplete information are available (e.g. [6]). Experiments with human subjects operating in an unknown maze [4, 5] suggest that humans might be able to handle this case as well.
The immediate problem is to determine if the resulting three dimensional C-space will still be as helpful to the human in performing the task as the two dimensional case was. Also necessary will be: algorithms for computing C-space for a multi-link arm; algorithms for collision analysis in 3D space; procedures for C-space visualization, and for arm motion control in W-space. These are likely to raise issues of computational complexity and real-time control.