03.11.2014 Views

18th annual conference on manual control.pdf - Acgsc.org

18th annual conference on manual control.pdf - Acgsc.org

18th annual conference on manual control.pdf - Acgsc.org

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

esulting angular positi<strong>on</strong> and stretch-evoked electromvograohic<br />

activity was m<strong>on</strong>itored. All experiments were performed under<br />

the c<strong>on</strong>trol of a real time COmDutinq system.<br />

Sinusoidal Oscillati<strong>on</strong>: The methods used and the oarameter<br />

estimati<strong>on</strong> techniques have been described by Agarwal and<br />

Gottlieb (1977a, 1977b) and Gottlieb and Agarwal (1978).<br />

Briefly, sinusoidal torques were applied to the wrist or ankle,<br />

and the resulting joint angles were measured. The ratio of<br />

the peak-to-peak joint angle to peak-to-peak torque amplitudes<br />

are computed for each of several frequencies. The sec<strong>on</strong>d<br />

order fit is computed to minimize the followin_ error<br />

criteri<strong>on</strong>:<br />

Error = _/ [ log (\Measured Model compliance_12<br />

complianco/.a<br />

In the wrist experiments a sliqht tapering of inout amolitudes<br />

was introduced to prevent excessive excursi<strong>on</strong> of the 'wrist at<br />

the lowest frequencies. Data collected in this way is<br />

qualitatively similar to data collected without the taoer,<br />

except that the wrist appears slightly stiffer at low<br />

frequencies than it ac_ually is.<br />

Step Tor_lue Perturbati<strong>on</strong>: The motor apnlied <strong>on</strong>e-sec<strong>on</strong>d<br />

durati<strong>on</strong> steps of torque to flex or extend the joint. These<br />

steps were typically randomized in amplitude and interstep<br />

interval. Subjects were instructed to either do nothinq when<br />

the step was oercieved, or react to restore the joint to its'<br />

starting positi<strong>on</strong>. Resp<strong>on</strong>ses were grouoed by perturbati<strong>on</strong><br />

amplitude and averaged.<br />

RESULTS<br />

Normal Compliance Measurements: Compliance measurements were<br />

made using sinusoidal torque inputs over a given frequency<br />

range (@.5 to i_.@ hz for the wrist, 3.Z to 12._ hz for the<br />

ankle). These measurements typically showed an underdamped<br />

sec<strong>on</strong>d order resp<strong>on</strong>se. Compliance curves for the wrist were<br />

c<strong>on</strong>structed at three different input amplitudes using five<br />

different bias torques, and are given xn fiqure 2. The<br />

parameters J, B, and K for these curves are shown in table i.<br />

The compliance measurement was very sensitive to input<br />

amplitude, and showed large decreases in K, the stiffness, as<br />

larger amplitude sinusoidal torques were used. The compliance<br />

measurements were also sensitive to bias :torque, showinq<br />

increases in K with increasing bias torques. Less pr<strong>on</strong>ounced<br />

but c<strong>on</strong>sistent changes were seen in B, the viscosity term. The<br />

81

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!