CS 530P Sensorimotor Computation
INSTRUCTOR: Dinesh K. Pai
(
)
Department of Computer Science, UBC
2010W: Term 1 (Sep-Dec 2010)
Lectures: TR 2:00-3:30 ICCS 206
News
The class room has been changed to ICCS 206
The first class is on Sep 9, following department policy.
Course Outline
This course provides a self-contained introduction to computational
models of sensorimotor biology, i.e., how the central nervous system
(CNS), muscles, and sensory organs work together to accomplish amazingly
effective movements that are still unmatched by robots. The goal is to
develop a constructive understanding of how these biological systems
work as machines, so that we can build predictive computer models and
simulations. Such an understanding is important not only for
biomedical applications but also for producing the next generation of
robots and computer animation.
The course will alternate between describing the physiology of
sensorimotor systems and developing computational models of such
systems. The latter will form a short introduction to physically
based modeling that is also useful for robotics and computer
graphics. We will use two concrete examples throughout the
course: eye movements and manipulation with hands.
No textbook is required. This year I will provide preliminary chapters
of a book I'm writing on this topic. Reference material will be
available in the reading room.
To make the course accessible to students with diverse backgrounds,
there will be a lot of flexibility in the choice of course projects to
suit individual needs. Email me if you have any questions.
There will be several synergistic activities in connection with
a Major Thematic
Grant from Peter Wall Institute on Sensorimotor Computation,
including a distinguished lecture series.
Topics:
- A first look at the human eye. Functional anatomy. Types of eye
movements. How the brain controls eye movements.
- Review of dynamical systems, using the control of saccadic eye
movements in the plane as an example. Differential equations. State
space. Stability. Control. Estimation. Numerical integration.
- Foundations of sensorimotor neurobiology.
- Computing on the cell membrane. Ion channels. Communication with
spikes. Synaptic transmission. Simplified neuron models.
- Muscle, the engine of movement. Sarcomere. Muscle fiber. Muscle
architecture. Motor units and recruitment. Mechanical transmission.
- Sensing. Signal transduction. Retina. Vestibular system. Muscle
spindle and GTO. Tactile mechanoreceptors.
- Eye movement in 3D. Listing's Law. Elementary differential
geometry. Classical mechanics on SE(3). Constraints. Principles of
Gauss and Hamilton.
- Modeling and simulation of biomechanisms. Multibody
dynamics. Deformable objects in one and more dimensions. Cables and
tensegrities. Case studies: muscle pulleys and eye movement; the
finger extensor mechanism.
- Contact with the external world. Friction. Convex analysis and
optimization.
- Biological Motor Control. Cerebellum, internal models, and
adaptive control. Motor learning. Psychophysical evidence for
optimality in biological motor control. Optimal feedback control and
the Hamilton-Jacobi-Bellman equation. Model reduction and muscle
synergies.
Evaluation:
The course grade will be based on
(1) Class participation. Everyone is expected to read assigned reading material, and each student will be expected to make one presentation (15%)
(2) Three assignments, with several questions requiring the use of Matlab (40%)
(3) One course project. This could involve either a critical review
of a topic or an implementation (45%).
Dinesh K. Pai