R.B. Gillespie

Shared control between human and machine: haptic display of automation during manual control of vehicle heading

In this paper, a paradigm for shared control is described in which a machines manual control interface is motorized to allow a human and an automatic controller to simultaneously exert control. The manual interface becomes a haptic display, relaying information to the human about the intentions of the automatic controller while retaining its role as a manual control interface. The human may express his control intentions in a way that either overrides the automation or conforms to it. The automatic controller, by design, aims to create images in the mind of the human of fixtures in the shared workspace that can be incorporated into efficient task completion strategies. The fixtures are animated under the guidance of an algorithm designed to automate part of the human/machine task. Results are presented from 2 experiments in which 11 subjects completed a path following task using a motorized steering wheel on a fixed-base driving simulator. These results indicate that the haptic assist through the steering wheel improves lane keeping by at least 30% reduces visual demand by 29% (p<0.000!) and improves reaction time by 18 ms (p=0.0009).

A Fundamental Tradeoff Between Performance and Sensitivity Within Haptic Rendering

In this paper, we show that for haptic rendering using position feedback, the structure of the feedback loop imposes a fundamental tradeoff between accurate rendering of virtual environments and sensitivity of closed-loop responses to hardware variations and uncertainty. Due to this tradeoff, any feedback design that achieves high-fidelity rendering incurs a quantifiable cost in terms of sensitivity. Analysis of the tradeoff reveals certain combinations of virtual environment and haptic device dynamics for which performance is achieved only by accepting very poor sensitivity. This analysis may be used to show that certain design specifications are infeasible and may guide the choice of hardware to mitigate the tradeoff severity. We illustrate the predicted consequences of the tradeoff with an experimental study.

Haptic feedback and human performance in a dynamic task

This study explores the effects of haptic feedback on performance and learning by human subjects executing a dynamic task. We present the results of experiments involving the control of a ball and beam. Human subjects perform position targeting of the ball through hand operation of the beam angle. In our dynamic analysis we discuss how this prototype task may be used to test the efficacy of various haptic feedback conditions. We obtain results for two conditions of haptic feedback, produced using two ball sizes, and apply various metrics to analyze performance. We also examine ordering effects that occur in the presentation of these haptic conditions. Our analysis and experimental findings indicate that the performance of a dynamic task is governed by the complexity of system dynamics and the magnitude of haptic feedback. Our results provide modest support to recommend exposure to a more complex, higher force-feedback task prior to the execution of a simpler lower feedback task.

Extremal distance maintenance for parametric curves and surfaces

A new extremal distance tracking algorithm is presented for parametric curves and surfaces undergoing rigid body motion. The essentially geometric extremization problem is transformed into a dynamical control problem by differentiating with respect to time. Extremization is then solved with the design of a stabilizing controller. We use a feedback linearizing controller. The controller simultaneously accounts for the surface shape and motion while asymptotically achieving (and maintaining) the extremal pair. Thus collision detection takes place in a framework fully analogous to the framework used for the simulation of dynamical response.

Human Adaptation to Interaction Forces in Visuo-Motor Coordination

We tested whether humans can learn to sense and compensate for interaction forces in contact tasks. Many tasks, such as use of hand tools, involve significant interaction forces between hand and environment. One control strategy would be to use high hand impedance to reduce sensitivity to these forces. But an alternative would be to learn feedback compensation for the extrinsic dynamics and associated interaction forces, with the potential for lower control effort. We observed subjects as they learned control of a ball-and-beam system, a visuo-motor task where the goal was to quickly position a ball rolling atop a rotating beam, through manual rotation of the beam alone. We devised a ball-and-beam apparatus that could be operated in a real mode, where a physical ball was present; or in a virtual training mode, where the balls dynamics were simulated in real time. The apparatus presented the same visual feedback in all cases, and optionally produced haptic feedback of the interaction forces associated with the balls motion. Two healthy adult subject groups, vision-only and vision-haptics (each n=10), both trained for 80 trials on the simulated system, and then were evaluated on the real system to test for skill transfer effects. If humans incorporate interaction forces in their learning, the vision-haptics group would be expected to exhibit a smoother transfer, as quantified by changes in completion time of a ball-positioning task. During training, both groups adapted well to the task, with reductions of 64%-70% in completion time. At skill transfer to the real system, the vision-only group had a significant 35% increase in completion time (p<0.05). There was no significant change in the vision-haptics group, indicating that subjects had learned to compensate for interaction forces. These forces could potentially be incorporated in virtual environments to assist with motor training or rehabilitation

Automated Characterization and Compensation for a Compliant Mechanism Haptic Device

Compliant mechanisms and voice coil motors can be used in haptic device designs to eliminate bearings and achieve smooth friction-free motion. The accompanying return-to-center behavior can be compensated using feedforward control if a suitable multidimensional stiffness model is available. In this paper we introduce a method for automatic self-characterization and compensation, and apply it to a planar haptic interface that features a five-bar compliant mechanism. We show how actuators and position sensors already native to typical impedance-type haptic devices can readily accommodate stiffness compensation. Although a portion of the motor torque is consumed in compensation, the device achieves smooth friction-free articulation with simple, low tolerance, and economic components. Empirical models built on self-characterization data are compared to standard empirical and analytical models. We produce a model by self-characterization that requires no inversion and is directly useable for compensation. Although our prototype compliant mechanism, which we fabricated in plastic using fused deposition modeling, exhibited hysteresis (which we did not compensate), the return-to-center behavior was reliably reduced by over 95% with feedforward compensation based on the self-characterized model.

Feedback-stabilized minimum distance maintenance for convex parametric surfaces

A new minimum-distance tracking algorithm is presented for moving convex bodies represented using tiled-together parametric surface patches. The algorithm is formulated by differentiating the geometric minimization problem with respect to time. This produces a hybrid dynamical system that incorporates dependence on rigid body motion, surface shape, and surface boundary interconnectedness. The minimum distance between a pair of previously identified closest features is found by feedback stabilizing the dynamical equations and numerically solving the resulting closed-loop system equations. Maintenance of the minimum distance and the associated closest points during motion is achieved through the action of a feedforward controller and a switching algorithm. The feedforward controller simultaneously accounts for surface shape and motion while the switching controller triggers updates to the extremal feature pair when extremal points on one body cross between Voronoi regions of the other body. In contrast to previously available minimum distance determination algorithms, attractive properties of the new algorithm include a means of determining the highest gain K that maintains stability under a given discretization scheme and a large and easily characterized basin of attraction of the stabilized closest points. These properties may be used to achieve higher computational efficiency. Simulation results are presented for various planar and spatial systems composed of a body and point or composed of two bodies.

Haptic feedback improves manual excitation of a sprung mass

In this paper, we present an experiment in which human subjects were asked to manually excite a virtual sprung mass into resonance under various feedback conditions: visual, haptic or visual and haptic combined. We are interested in comparing the value of these feedback conditions in terms of their influence on the achievable performance in a dynamic task such as exciting a resonant mechanical system. From our human subject experiment (n=10), we found that with haptic feedback alone, subjects successfully excited the sprung mass into resonance. For the particular case of /spl omega//sub n/ = 7 rad/s, subjects demonstrated significantly larger differences between the observed and expected frequency distribution under vision-only (paired t-test: p=0.034) and haptics-only feedback conditions (paired t-test: p=0.021), as compared to combined vision with haptic feedback. Variability of key marker locations of input behavior were also significantly lower with both feedback channels than with either alone (paired t-tests: p<.0002). Our results show that haptic feedback can augment vision to produce significant improvements in the control of a dynamic system.

A closest point algorithm for parametric surfaces with global uniform asymptotic stability

We present an algorithm that determines the point on a convex parametric surface patch that is closest to a given (possibly moving) point. Any initial point belonging to the surface patch converges to the (possibly moving) closest point without ever leaving the patch. Thus the algorithm renders the patch invariant and is globally uniformly asymptotically stable. The algorithm is based on a control problem formulation and solution via a switching controller and common control Lyapunov function. Analytic limits of performance are available, delineating values for control gains needed to out-run motion (and shape) and preserve convergence under discretization. Together with a top-level Voronoi diagram-based switching algorithm, the closest point algorithm treats parametric models formed by tiling together convex surface patches. Simulation results are used to demonstrate invariance of the surface patch, global convergence, limits of performance, relationships between low-level and top-level switching, and a comparison to competing Newton-iteration based methods.

Haptic rendering of parametric surfaces using a feedback stabilized extremal distance tracking algorithm

An extremal distance tracking algorithm is presented for convex parametric curves and surfaces undergoing rigid body motion. The geometric extremization problem is differentiated with respect to time to produce a dynamical system that incorporates dependence on both surface shape and rigid body motion. Extremization then takes place by integrating these dynamical equations, but with a feedback controller in place to stabilize the solution. A controller design using feedback linearization is developed that simultaneously accounts for surface shape and motion while asymptotically achieving (and maintaining) the extremal pair. Collision detection then takes place in a framework fully analogous to that used for multibody simulation. Local stability results are extended to provide global stability for body shapes composed of pieced-together convex parametric surface patches using a switching algorithm.