Postgraduate Course: Machine Learning and Sensorimotor Control (INFR11014)
Course Outline
| School | School of Informatics |
College | College of Science and Engineering |
| Course type | Standard |
Availability | Available to all students |
| Credit level (Normal year taken) | SCQF Level 11 (Postgraduate) |
Credits | 10 |
| Home subject area | Informatics |
Other subject area | None |
| Course website |
http://www.inf.ed.ac.uk/teaching/courses/mlsc |
Taught in Gaelic? | No |
| Course description | Control of complex, compliant, multi degree of freedom sensorimotor systems like humanoid robots or autonomous vehicles have been pushing the limits of traditional control theoretic methods. This course aims at introducing adaptive and learning control as a viable alternative. The course will take the students through various aspects involved in motor planning, control, estimation, prediction and learning with an emphasis on the computational perspective. We will learn about statistical machine learning tools and methodologies particularly geared towards problems of real-time, online learning for sensorimotor control. Issues and possible approaches for multimodal sensor integration, sensorimotor transformations and learning in high dimensions will be discussed. This will be put in context through exposure to topics in human motor control, experimental paradigms and the use of computational methods in understanding biological sensorimotor mechanisms. |
Information for Visiting Students
| Pre-requisites | None |
| Displayed in Visiting Students Prospectus? | No |
Course Delivery Information
| Not being delivered |
Summary of Intended Learning Outcomes
1 - Describe the components of 'traditional' model based control and critically assess it's limitations in the real-time control of compliant, high dimensional sensorimotor systems.
2 - Design and evaluate experimental paradigms to test various biological control hypotheses and identify potential ways of reducing control complexity.
3 - Apply the machine learning tools and algorithms learnt in the module to design an efficient adaptive (learning) control scheme for a given real world control problem.
4 - Implement a basic ?model-based= control schema with learning in MATLAB/C using dimensionality reduction techniques.
5 - Carry out benchmark comparisons against the state of the art learning methods and optimization/planning strategies. |
Assessment Information
Written Examination 50
Assessed Assignments 35
Oral Presentations 15
Assessment
There will be two coursework assignments accounting for 35% of the course marks. One of these might involve a group mini-project involving implementation of learning control on an artificial simulated plant or practical implementation of sensor integration strategies. 15% of the marks will be based on a paper presentation from a list of human motor control papers.
If delivered in semester 1, this course will have an option for semester 1 only visiting undergraduate students, providing assessment prior to the end of the calendar year. |
Special Arrangements
| None |
Additional Information
| Academic description |
Not entered |
| Syllabus |
Fundamental Control Theory
*Classical control: PID, model based vs direct control
*Limitations of traditional control
Machine Learning Tools for Learning Control
*Dimensionality reduction
*Locally Weighted Learning and other nonparametric methods
*Real time / Online learning and Distal Learning
Adaptive Control and Learning Control
*Fundamentals: Trajectory planning, Inverse Kinematics, Inverse Dynamics
*Multiple paired forward inverse models (MPFIM)
*Synergistic control and co-activation
*Coordinate transformations: body centric, retinotopic, object centric etc.
Predictive control: Kalman filtering and Particle filters
Dynamical Systems as movement policies : extracting, learning and tuning primitives
Bilological/Human Motor Control
*Force field hypothesis, equilibrium point hypothesis
*Internal models and cerebellum
*Tuning curves and force adaptation
Sensorimotor Integration
*Bayesian cue integration
*Sensor fusion- models and effects
*Explaining away and cue reliability based integration
Relevant QAA Computing Curriculum Sections: Artificial Intelligence, Intelligent Information Systems Technologies |
| Transferable skills |
Not entered |
| Reading list |
* J-J. Slotine and W. Li (1991), Applied Nonlinear Control, Prentice-Hall.
* T. Hastie, R. Tibshirani, and J. Friedman. The Elements of Statistical Learning. Springer Series in Statistics. Springer Verlag, Basel, 2001.
* Latash, M. (1993). Control of Human Movement. Champaign, IL: Human Kinetics Publ.
* Current Opinion in Neurobiology (1999) vol. 9, Special issue on Motor Control, Elsevie
* Hastie, T., Tibshirani, R. (1995): Generalized additive models. Chapman and Hall, London. |
| Study Abroad |
Not entered |
| Study Pattern |
Lectures 20
Tutorials 0
Timetabled Laboratories 0
Non-timetabled assessed assignments 40
Private Study/Other 40
Total 100 |
| Keywords | Not entered |
Contacts
| Course organiser | Dr Michael Rovatsos
Tel: (0131 6)51 3263
Email: mrovatso@inf.ed.ac.uk |
Course secretary | Miss Gillian Bell
Tel: (0131 6)50 2692
Email: Gillian.Watt@ed.ac.uk |
|
|
University Homepage