Engineering topics
Publications and source records attributed to Hill, J. W..
Task board constructed to facilitate time-and-motion studies for remote manipulators. Apparatus equipped with holes, objects of various shapes to be grasped and sensors with switches to indicate contact. Useful in industrial robots programmed to assemble parts.
Controller for remote manipulators gives user "feel" for forces required to lift, slide, turn, and otherwise handle objects. Because operator experiences sensations similar to those he would perceive if he handled objects directly, he needs much less skill and training for manipulator than for one with force feedback. It was developed to handle hazardous materials, such as radioactive substances, explosives, or corrosive chemicals. Other possible uses include tracking moving objects, vehicle control, and human interaction with computers (for example, via three dimensional display of computer model).
The use of time and motion study methods to evaluate force feedback in remote manipulation tasks are described. Several systems of time measurement derived for industrial workers were studied and adapted for manipulator use. A task board incorporating a set of basic motions was designed and built. Results obtained from two subjects in three manipulation situations for each are reported: a force-reflective manipulator, a unilateral manipulator, and the unaided human hand. The results indicate that: (1) a time-and-motion study techniques are applicable to manipulation; and that (2) force feedback facilitates some motions (notably fitting), but not others (such as positioning).
A description is given of part of a continuing effort both to develop models for and to augment the performance of humans controlling remote manipulators. The project plan calls for the performance of several standard tasks with a number of different manipulators, controls, and viewing conditions, using an automated performance measuring system; in addition, the project plan calls for the development of a force-reflecting joystick and supervisory display system.
Sensor employs triangulation technique to locate objects that lie in intersections of four emitted beams of light and fields of view of four phototransistors. Signals from individual phototransistors are filtered and identified by individual synchronous detectors, one for each beam-intersection point.
The results are described from two manipulators on a peg-in-hole task, which is part of a continued effort to develop models for human performance with remote manipulators. Task difficulty is varied by changing the diameter of the peg to be inserted in a 50 mm diameter hole. An automatic measuring system records the distance between the tool being held by the manipulator and the receptacle into which it is to be inserted. The data from repeated insertions are processed by computer to determine task times, accumulated distances, and trajectories. Experiments with both the MA-11 cable-connected master-slave manipulator common to hot cell work and the MA-23 servo-controlled manipulator (with and without force feedback) are described. Comparison of these results with previous results of the Ames Manipulator shows that force feedback provides a consistent advantage.
Real-time performance data was collected during a pick-up task carried out with a Rancho master-slave manipulator using a minicomputer-based data taker. In addition to the usual task-time measurements, computer algorithms to integrate the energy consumed and to count and time the number of moves were implemented. In addition to these measures, several derived measures such as the fraction of time moving (MRATIO) and mean time per move (MBAR) were obtained in an off-line analysis. Preliminary results of the time delay experiment indicate that two new measures, MRATIO and MBAR, are almost an order of magnitude more sensitive than task time, the conventional measure, in determining performance changes with transmission delays in the range from 0.0 to 1.0s.
Modeling of human performance in remote manipulation tasks is reported by automated procedures using computers to analyze and count motions during a manipulation task. Performance is monitored by an on-line computer capable of measuring the joint angles of both master and slave and in some cases the trajectory and velocity of the hand itself. In this way the operator's strategies with different transmission delays, displays, tasks, and manipulators can be analyzed in detail for comparison. Some progress is described in obtaining a set of standard tasks and difficulty measures for evaluating manipulator performance.
Human performance measurement techniques for remote manipulation tasks and remote sensing techniques for manipulators are described for common manipulation tasks, performance is monitored by means of an on-line computer capable of measuring the joint angles of both master and slave arms as a function of time. The computer programs allow measurements of the operator's strategy and physical quantities such as task time and power consumed. The results are printed out after a test run to compare different experimental conditions. For tracking tasks, we describe a method of displaying errors in three dimensions and measuring the end-effector position in three dimensions.
Remote manipulator control considered is based on several levels of automatic supervision which derives manipulator commands from an analysis of sensor states and task requirements. Principle sensors are manipulator joint position, tactile, and currents. The tactile sensor states can be displayed visually in perspective or replicated in the operator's control handle of perceived by the automatic supervisor. Studies are reported on control organization, operator performance and system performance measures. Unusual hardware and software details are described.
Description of the equipment employed and results obtained in experiments with tactile feedback and different levels of automatic control. In the experiments described tactile feedback was investigated by incorporating a touch sensing and touch display system into a teleoperator, while the levels of automatic control were investigated by incorporating supervisory control features in the teleoperator control system. In particular, a hand contact system which senses and reproduces to the operator the contact between the end-effector and the object being touched or manipulated is described, as well as a jaw contact system which senses and reproduces to the operator the shape and location of the object held in the remote jaws, and an arm control system consisting of a control station where the operator controls the motion of the arm by transmitting commands, a remote station that accepts the commands and uses them, and a communications link that limits information flow. In addition, an algorithmic language for remote manipulation is described, and the desired features that an automatic arm controller should possess are reviewed.
A hand/touch sensing system is described that, when mounted on a position-controlled manipulator, greatly expands the kinds of automated manipulation tasks that can be undertaken. Because of the variety of coordinate conversions, control equations, and completion criteria, control is necessarily dependent upon a small digital computer. The sensing system is designed both to be rugged and to sense the necessary touch and force information required to execute a wide range of manipulation tasks. The system consists of a six-axis wrist sensor, external touch sensors, and a pair of matrix jaw sensors. Details of the construction of the particular sensors, the integration of the end effector into the sensor system, and the control algorithms for using the sensor outputs to perform manipulation tasks automatically are discussed.
A computer-assisted teleoperator control system for making comparative performance evaluations is described. A local and a remote control station, each with decision-making capability, communicate with each other through a simulated time delay. Supervisory control at three increasingly automatic levels is possible. The highest level of programmed control is facilitated through the ARM language which was developed to permit easily readable program manuscripts to be written and assembled into programs of motions by novice programmers. Experimental results show the advantage of this form of supervisory control with both direct and delayed (3 sec) manipulation tasks. In addition, two systems to measure and reproduce force distributions have been designed. One system reproduces contact on the external surfaces of the remote hand using 21 airjet simulators. Another system reproduces the shape of the contact between object and jaws using 288 piezoelectric (bimorph) stimulators.
Computer assisted teleoperator control system for making comparative performance evaluations of tactile perception
Performance differences between tactile and visual localization and temporal ordering ability, using sequential presentation of high rate point stimuli
Experimental tests to measure short term memory characteristics of visual and tactile perception related to teleoperator systems
Human perception of multiple-point tactile and visual stimuli
Small eye movements measurement and retinal image stabilization based on tracking edge of blood vessel in optic disk