View Full Version : FWD from NeuroMotor-Control list

H.j. Woltring, Fax/tel +31.40.413 744
09-08-1992, 02:23 AM
Date: Tue, 8 Sep 92 08:53:17 EDT
From: reza@ai.mit.edu
Subject: Dissertation Defense Announcement

It's my pleasure to announce that Ernie Fasse will be defending
his dissertation this Friday, September 11, at 3:30 pm, in
room 3-343 at MIT. The abstract follows.

For those interested in getting a copy of the dissertation, please
send a note to me and I will see what can be arranged. There
might be a small charge for copying, binding, and shipping.

Reza Shadmehr, Ph.D.
Dept. of Brain and Cognitive Sciences, M. I. T.


Ph.D. Dissertation Defense


Ernest Dean Fasse
3:30 p.m., Friday, September 11, 1992
Room 3-343, M. I. T.

Robots and humans use mechanical sensory information from the arm in control of
the arm as well as in perception of mechanical properties of the arm and
external objects. This thesis looks at the use and representation of sensory
information in control and perception of robotic and human arms. One observa-
tion is that humans are capable of purposeful movement in the absence of sen-
sory information whereas most robots are not. This difference in robustness to
sensory deprivation is shown to be attributable to a difference in the under-
lying stability properties of the actuators. A preliminary actuator design is
presented that is robust to deprivation of sensory information.

The issue of representation of information is important in any discussion of
computation. A series of experiments were performed to look for evidence of
structure in the way humans represent mechanical information. The principal
hypothesis tested was that human perception of mechanical qualities is dis-
torted in a locally linear, metrically consistent manner, as is commonly
assumed in robotic control schemes. Five perception experiments and two motor
control experiments were performed. In the perception experiments subjects
made judgements about mechanical properties of objects simulated by a planar
robot. In the first motor control experiment subjects drew circles. In the
second experiment subjects interacted with circular constraints simulated by
the robot.

The principal hypothesis is shown to be false. The perceptual and motor beha-
vior observed in the experiments could not be explained by a single, unifying,
locally linear metric. Two positive results of the experiments are: (1) The
so-called tangential-radial illusion, where radial distances are perceived to
be longer than equal tangential distances, is misnamed. Although human per-
ception of length is distorted on the order of 20% and repeatable among sub-
jects, there is nothing special about the tangential and radial directions.
(2) The shapes drawn in the circle drawing experiment were consistent with
those predicted by the length perception experiment. This suggests that both
behaviors are manifestations of a common, underlying phenomenon.

Thesis Committee:
Prof. Neville Hogan, Dept. of Mechanical Engineering (Chairman)
Prof. William K. Durfee, Dept. of Mechanical Engineering
Dr. Ferdinando Mussa-Ivaldi, Dept. of Brain and Cognitive Sciences
Prof. Jean-Jacques Slotine, Dept. of Mechanical Engineering