Model-Based Cancellation of Biodynamic Feedthrough Using a Force-Reflecting Joystick

[+] Author and Article Information
R. Brent Gillespie

Szabolcs Sövényi

Department of Mechanical Engineering,  University of Michigan, 2350 Hayward St., Ann Arbor, MI 48109ssovenyi@umich.edu

J. Dyn. Sys., Meas., Control 128(1), 94-103 (Sep 23, 2005) (10 pages) doi:10.1115/1.2168480 History: Received April 04, 2005; Revised September 23, 2005

Manual control performance on-board a moving vehicle is often impeded by biodynamic feedthrough—the effects of vehicle motion feeding through the operator’s body to produce unintended forces on the control interface. In this paper, we propose and experimentally test a model-based controller that acts through a motorized manual interface to cancel the effects of biodynamic feedthrough. The cancellation controller is based on characterization data collected using an accelerometer on the vehicle and a force sensor embedded in the manual interface and a protocol under which the manual interface is temporarily immobilized while in the grip of the operator. The biodynamic model fit to the data is based in turn on a carefully constructed model of the coupled vehicle-operator system. The impact of biodynamic feedthrough and the ability of the model-based controller to cancel its effects were estimated through an experiment in which 12 human subjects used a joystick to carry out a pursuit tracking task on-board a single-axis motion platform. Cancellation controllers derived from biodynamic models fit individually to each subject significantly improved pursuit tracking performance, as evidenced by a 27% reduction in root-mean-square tracking error, a 32% improvement in time-on-target, and an increase in crossover frequency from 0.11 to 0.15 Hz.

Copyright © 2006 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Figure 1

A human operator seated on a single-axis motion platform uses a joystick to cause a cursor on the screen to track a target that moves in an unpredictable fashion. The translational axis of the motion platform is perpendicular to the rotational axis of the joystick; thus both the platform and hand motions are in the lateral direction.

Grahic Jump Location
Figure 2

The human operator is modeled as a two-input, two-output system in which the input velocity ẋv and output force fs comprising port 1 capture the interaction between the trunk and the vehicle seat, while the output force fb and input velocity ẋj comprising port 2 describe the interaction between the hand and the joystick. The four impedances capture the input-output maps of the two-port. The transfer function T describes how the operator responds to the visually observed difference between the reference signal xr and the plant output xp by imposing a force ft on the joystick J. The force fb enters the tracking loop as a disturbance and models the biodynamic response of the human operator to the joystick angular velocity ẋj and the vehicle velocity ẋv.

Grahic Jump Location
Figure 3

In this block diagram, biodynamic feedthrough can be recognized as a pathway by which vehicle acceleration ẍv enters the tracking loop as a disturbance. This block diagram follows from that in Fig. 2 after removing Z12 and Z11 under the assumption that the vehicle acts as a motion source and after defining H≡Z21∕s−mj and moving Z22 into position as a feedback loop around the joystick J.

Grahic Jump Location
Figure 4

The frequency response of the force fb′ to the excitation ẍv during the system identification test is shown for one subject. The Bode plot of the model Ĥ fit to the experimental data is shown in a continuous line.

Grahic Jump Location
Figure 5

A generic nonlinear system expressed as the sum of a describing functionT and a remnant n(t)

Grahic Jump Location
Figure 6

The single-axis motion platform is driven by DC motor and ballscrew and features a joystick accessible to the right hand of a seated operator

Grahic Jump Location
Figure 7

System identification results for 12 subjects. The models show similar trends, but it appears that hey cannot be substituted with a single, average model. We therefore propose the construction of a separate controller for each individual.

Grahic Jump Location
Figure 8

20s of the reference xr and plant output xp signals are shown for a typical subject under the three experimental conditions: (a) stationary platform, (b) moving platform without compensation, and (c) moving platform with compensation

Grahic Jump Location
Figure 9

rms error averages with 10s moving time windows under the three test conditions

Grahic Jump Location
Figure 10

Dwell ratios averages with 10s moving time windows under the three test conditions

Grahic Jump Location
Figure 11

Boxplot of rms error values across the 12 subjects under the three test conditions

Grahic Jump Location
Figure 12

Boxplot of Dwell ratios across the 12 subjects under the three test conditions

Grahic Jump Location
Figure 13

Open loop transfer function of tracking under the three test conditions

Grahic Jump Location
Figure 14

Boxplot of crossover frequencies across the 12 subjects under the three test conditions




Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In