Last modified:




Biophysics 586N: Introduction to Neuromorphic Engineering

COURSE DESCRIPTION 2008

This course will introduce neuromorphic engineering as a tool to study neurobiological and biomechanical problems. We will participate in a hands-on experience building neurons, synapses and simple neural networks from analog circuits which will subsequently control servo-motors arranged to model after the body shapes and movement styles of insects, crustaceans and other invertebrates. The class will meet twice a week for a lecture/discussion and a lab. During the lecture/discussion period (1 hour once a week) we will present and discuss material related to neuronal pattern generators, animal movement, robotic models of animals and robotics application in research and industry. In lab (3 hours once a week) each of us will build a simple neuromorphic robot of their own and subsequently attempt to use it to investigate a specific research problem in neuroscience or biomechanics. The study will culminate in writing of a scientific paper where each student will discuss the resulting robot with appropriate literature review and relating their work to the field. The goal is that the undergraduate students could be encouraged to submit their papers for publication in the Undergraduate Neuroscience Journal and the graduate students would have at their disposal a strong example of a well-written article to serve them in their later endeavors.

This course is intended for upper-level undergraduates as well as graduate students who are studying neuroscience, biophysics and computer programming. It is expected that the biology students will have little, if any, familiarity with engineering and computer programming while the engineering students will have little, if any, familiarity with biology and neuroscience. Thus, the final grade will not necessarily correlate with the quality of the final product, a robot, but rather with the quality of the paper, which will reflect how well-rounded and integrated is the author's understanding of the current problems in neuromorphic engineering, and how much the author has improved their skills and knowledge outside their major field.
Alternatives to this basic format will be permitted. All students will be expected to make a simple CPG (minimum 2 neurons) on a breadboard, and thus a write-up of that will be required as part of methodology, results, and discussion. However, if you are not absolutely successful in programming the controller and/or in getting the robot to work, you will not be able to include that in the paper. Thus, you can expand the literature review from, say, 15 papers to 30 papers and emphasize the importance of the work you did accomplish, and its relevance to the state of the art in the field. This way you can still get a good grade without killing yourself. You will have considerable freedom in the choice of a project, thus I cannot design a single rigid set of guidelines. Suggested "fuzzy" guidelines are presented at the end of the syllabus, and we will be discussing options during the course of the semester as well..

One of the major challenges in this course will be designing your own project, based on some background you acquire in discussion section. To help you, I compiled a collection of tools and suggestions drawn from the field of entrepreneurial development. Finding new ideas, designing and implementing new projects is what entrepreneurs do for a living. Consider reading some of their suggestions - you may find them helpful, albeit in need of being adapted to our particular area. Just substitute the "what do we earn through this project?" with the "what do we learn through this research?".



MY GOALS FOR THE COURSE - what I hope you will get out of this:

MY INSTRUCTIONAL OBJECTIVES - what I will do for you:

What I will NOT do for you:



MATERIAL COVERED IN DISCUSSION AND LAB

By Week 4 each group of students will choose one well-studied animal (e.g. an insect, mollusk or crustacean) for which the CPG for movement (walking, crawling, swimming) has been discovered. Students will model their robot after that animal, implementing the CPG on the breadboard and connecting the motors and sensors so as to come as close as possible to the natural movement. After that, if there is time leftover in the semester, the circuit may be mounted on a platform bearing the motors and sensors, to create a finished-looking prototype. Time permitting, the robot will be used to investigate a research problem related to the field. I will be working on my own project in parallel with you, so you will have an example at all times if you need help or get stuck.

Each project will likely be very different from others. Thus, in this Syllabus, I can provide specific recommendations only at beginning stages: building a CPG on a breadboard, programming a PIC to generate spikes and control motors. After that, everything you do will be specific to your robot, and the syllabus entries are necessarily much fuzzier. Do ask for help or clarification, if you feel the need.

We will try to use the "two-cool", "idea-bounce", "the pitch" and "the idea draft" - the techniques borrowed from entrepreneurs - to help acquire a vision of the project and define achievable goals, as well as to learn how to present technical subjects to coworkers with a different background. Homeworks are designed to serve as steps for writing your paper.