Fundamentals of Neuromechanics
Course Flyer (pdf; 1.8kB)Class Photo (jpg; 100kB)
Student Evaluations of the Class: summary of written comments and numerical scores (BME section only)
When: Tuesdays and Thursdays, 5:30-7:20 PM
Where:
Technological Institute,
Room M128Biomedical Engineering course number: BMD_ENG495-0
Mechanical Engineering course number: MECH_ENG495-0
Instructor: Malcolm MacIver
What you will learn
- how to make a polygonal computer model of an animal
- how to estimate the sensory signals going to the brain during behavior using a polygonal model of the body, motion capture data, and models of the stimulus and receptor
- how to estimate muscle activation (EMG) based on motion capture data and a musculoskeletal model
- estimating the interaction between a mechanical device and the human body for neuroprosthetics
- the use of optimality theory in understanding animal behavior
- how biomechanics relates to the information needed to guide behavior
- key concepts of sensorimotor interaction in animals
- when and how to use robotic approaches for understanding sensorimotor function
Course description: The mechanical interaction of the body with its environment is a key player in nervous system function due to its influence on behavior. In this course, we will take a systems-level view of animals and explore relationships between behavior, biomechanics, and the nervous system through examples from the literature and through projects with two model systems: weakly electric fish, and simple human arm movements.
A significant component of the course will involve assignments using two different simulation environments: AnimalLab a computational neuroscience simulation environment that enables reconstruction of sensory signal input to the brain from motion capture data of natural behavior; and
AnyBody,
a computational biomechanics package that enables reconstruction of
muscle activation patterns from motion capture data.
Using AnimalLab, students will learn about the creation and manipulation of animal body models, sensor layouts, and motion patterns of an active sensing animal, and how animal movement relates to sensory input.
In one of the AnimalLab-related assignments, students will use the laser scanner in the Department of Mechanical Engineering to make a computer model of an animal body of their choice. We will use plastic models for the scan if needed. An advanced software package for manipulating scan data (
RapidForm) will be used for creation of the model from the scan data.
Using AnyBody, students will build up a model of the human arm, using supplied muscle and skeleton properties, and use this model to predict arm EMG signals based on a simple arm movement. They will then compare EMG measurements of this arm movement, collected in
Professor Eric Perreault's laboratory, to these predictions and explore the source of any
discrepancies between the model and measurements.
This module of the course was developed with Eric Perreault.
The issue of coupled biomechanical-machine models, an important topic in neuroprosthetics, will also be explored. We will use a previously developed AnyBody model of a person riding a bicycle for this exercise. Students will learn how to increase peddling muscle efficiency by attaching passive springs to the bicycle. This module of the course was developed in cooperation with
Professor John Rasmussen,
of the Department of Mechanical Engineering
in Aalborg Denmark, the lead developer of AnyBody.
Prerequisites: Some facility with programming in Matlab will be needed. This can be acquired through tutorials if you are computer-savvy. Contact instructor for guidance on this. A computer with Matlab, and three additional software packages (AnimalLab, AnyBody, RapidForm) will also be necessary; if the student does not have a suitable computer there are two clusters of student computers which will be made available with the software already installed.
Students will be automatically given accounts on
Depot, which will be used for distribution of
software, readings, and depositing of assignments.
SYLLABUS
2004.09.23/Week 01/Class 01: Plants vs animals; the origin of the nervous system; the multiple timescales of adaptation to environment; how not to study adaptive behavior; the Parable of the Chip - how to reverse engineer the nervous system; an overview of one particular model system (weakly electric fish)
Assignment 01: Directed Reading 01—Creating surface models of animal bodies: polygonal and generative models. Papers:
(1) MacIver, M. A., & Nelson, M. E. (2000). Body modeling and model-based tracking for neuroethology. Journal of Neuroscience Methods, 95(2), 133-143.
(2) Ramamoorthi, R., & Arvo, J. (1999). Creating Generative Models from Range Images. Paper presented at the SIGGRAPH 99.
2004.09.28/Week 02/Class 02: Assignment 01 Due.
An introduction to modeling animal body surfaces. How to use 3D digitizing systems to acquire surface data. Using optical, MRI, and contact digitizers. What are generative models, how do they differ from polygonal models, where they can help.
In class assignment: Decide on an object to be 3D scanned with the Laser Surveyor in the ME Rapid Prototyping Lab. Divide into four-person groups, nominate a representative, and have that person coordinate with the Laser Surveyor operator.
Assignment 02: Computation 01—Manipulating motion capture data, and applying DOF to a surface model of a weakly electric fish using AnimalLab.
2004.09.30/Week 02/Class 03: Assignment 02 Due. Review of generative modeling. Giving the body sensation. What sensing is, the kinds of energy utilized in sensing, how biological sensory arrays are structured. Active sensing. The concepts of sensing and motor volume, extending the concept of reachability into a biological domain. An introduction to how sensory input is reconstructed: an example using electric fish. Introduction to computing the motor volume, aka "reachable set" of an electric fish.
Assignment 03: Directed Reading 02—Giving the body sensation. Papers:
(1) Nelson, M. E., & MacIver, M. A. (1999). Prey capture in the weakly electric fish Apteronotus albifrons: Sensory acquisition strategies and electrosensory consequences. Journal of Experimental Biology, 202(10), 1195-1203.
(2) Terzopoulos, D., Rabie, T., & Grzeszczuk, R. (1997). Perception and learning in artificial animals. In Artificial life V: proceedings of the Fifth International Workshop on the Synthesis and Simulation of Living Systems (pp. 313-320). Nara, Japan: MIT Press.
Optional paper: Neumann, T. R. (2002). Behavior-Orientated Vision for Biomimetic Flight Control. 196-203.
2004.10.05/Week 03/Class 04: Assignment 03 Due. Each group leader updates on progress of Laser Surveyor and RapidForm work to make a 3D surface model of the group's selected object. Detailed discussion of how to compute the sensing volume and small-time motor volume. Computing the fractional overlap of the sensing and motor volumes. Matching of sensing and small-time motor volumes in active sensing systems.
Assignment 04: Directed Reading 03—Morphology and modularity. Papers:
(1) Carroll, S. B. (2001). Chance and necessity: the evolution of morphological complexity and diversity. Nature, 409(6823), 1102-1109.
(2) Thomas, R. D. K., Shearman, R. M., & Stewart, C. W. (2000). Evolutionary exploitation of design options by the first animals with hard skeletons. Science, 288(5469), 1239-1242.
(3) Hornby, G., Lipson, H., & Pollack, J. B. (2003). Generative Representations for the Automated Design of Modular Physical Robots. IEEE TRANSACTIONS ON ROBOTICS AND AUTOMATION, 19(4), 703-719.
Assignment 05: Computation 02—Laser scanning and using RapidForm to make a water-tight polygonal model of the group's chosen object, and importing this model into MATLAB.
2004.10.07/Week 03/Class 05: The different control regimes enabled by different sensing to small-time motor ratios. The difference between planning and control understood using the concept of equivalence classes of algorithmic convergence.
2004.10.12/Week 04/Class 06: Assignment 04 Due. Continuation of discussion of the interconnection between control, planning, and the relative sizes of the sensory and motor volumes. Introduction to modularity in animal body plans. Exploratory and performatory modes of behavior in animals.
2004.10.14/Week 04/Class 07: Assignment 05 Due. Discussion of Hod Lipson's paper on generative models for the evolution of robots and the evolution of evolvability. Nielsen's Trochae Theory for the genesis of the metazoans.
2004.10.19/Week 05/Class 08: Discussion of mistakes noticed in previous directed reading responses. Review of reconstruction of signal input to the body during natural behavior. Discussion of reconstruction of neural input to the brain during natural behavior. What this effort with electric fish has taught us.
Assignment 06: Reconstruction of the sensory input to an electric fish's brain during prey capture behavior, and running this reconstructed input through an accurate afferent model. Major subroutines in AnimalLab.
Assignment 07: Group presentations.
2004.10.21/Week 05/Class canceled.
2004.10.26/Week 06/Class 09. Two group presentations:
Team Doh presents Dickinson, M. H., Farley, C. T., Full, R. J., Koehl, M. A. R., Kram, R., & Lehman, S. (2000). How animals move: An integrative view. Science, 288(5463), 100-106.
Team Chief presents (1)Full, R. J., & Koditschek, D. E. (1999). Templates and anchors: Neuromechanical hypotheses of legged locomotion on land. Journal of Experimental Biology, (202): 3325-3332.
(2) Thomas L. Daniel, and Michael S. Tu (1999). Animal Movement, mechanical tuning, and coupled systems. Journal of Experimental Biology, (202): 3415-3421.
2004.10.28/Week 06/Class 10. Two group presentations:
Team Foot presents William J. Kargo, Frank Nelson and Lawrence C. Rome (2002). Jumping in frogs: assessing the design of the skeletal system by anatomically realistic modeling and forward dynamic simulation. Journal of Experimental Biology (205):1683-1702.
Team Obelix presents (1) Ekeberg, O., Lansner, A., & Grillner, S. (1995). The neural control of fish swimming studied through numerical simulations. Adaptive behavior, 3(4), 363-384.
(2)Grillner, S. (1996). Neural networks for vertebrate locomotion. Scientific American, 274, 64-89.
2004.11.02/Week 07/Class 11. The difference between when a signal can be detected on a set of afferents and when there is a behavioral response—why the reconstructed neural detection time differs from the measured behavioral response in the assignment. Different approaches to processing the signals across thousands of afferents: the receptive field. Basics of spiketrain analysis. Using optimal control theory to investigate animal behavior: An example using weakly electric fish.
2004.11.04/Week 07/Class 12. Energy speculators versus energy conservators and how this is reflected in the locomotory abilities of animals. Review of important points brought up in the four group presentations.
2004.11.09/Week 08. Class canceled.
2004.11.11/Week 08/Class 13. Discussion of the final project: Using the software tools taught during the course, design an animal. Focus on modeling sensory aspects during artificial behavior, or focus on modeling biomechanical aspects.
Assignment 08: Directed Reading 04. Chapter 1 of: Forster, E. (2003). Predicting muscle forces in the human lower limb during locomotion. The Medical Faculty, University of Ulm, Germany.
2004.11.16/Week 09/Class 14. Basic muscle physiology. Geometry and mechanical properties of muscle fibers.
2004.11.18/Week 09/Class 15. Biomechanics meet information theory: the new domain of infomechanics. Discussion of two papers:
(1) Weissburg, M. J., James, C. P., Smee, D. L., & Webster, D. R. (2003). Fluid mechanics produces conflicting constraints during olfactory navigation of blue crabs, Callinectes sapidus. Journal of Experimental Biology, 206(1), 171-180.
(2) Laughlin, S. B. (2001). Energy as a constraint on the coding and processing of sensory information. Current Opinion in Neurobiology, 11(4), 475-480.
2004.11.23/Week 10/Class 16. Additional topics in muscle physiology. Discussion of neuromusculoskeletal muscles.
Assignment 09: Directed Reading 05. Paper:
Crago, P. E.. (2000). Creating Neuromusculoskeletal Models, Chapter 8 of Biomechanics and Neural Control of Posture and Movement by Jack M. Winters.
2004.11.25/Week 10: Class canceled (Thanksgiving).
2004.11.30/Week 11/Class 17. Assignment 09 Due. Optimality principles in sensorimotor control. Discussion of: Optimality principles in sensorimotor control by Emanuel Todorov (2004). Nature Neuroscience 9(7): 907-915. Open-loop and closed-loop control laws. How to determine whether a behavior involves sensory feedback: an example from electric fish, and from echolocating bats. Bat calls and the coupling of bat locomotion to bat signal generation. Bat prey capture reconstructions with echolocation beams from Moss lab.
2004.12.02/Week 11/Class 18. The open-loop/closed loop continuum. Why some animals have to use a stop-and-go strategy due to inadequate reafference suppression. Using moth evasion tactics to learn about limits to bat motion control. Review of major concepts discussed throughout the class. The neuromechanical manifesto.
