Discussion August 28: Introduction to neuromorphic engineering and its applications in science and industry. Several examples of research application of neuromorphic engineering will be considered, based on the research articles. I recommend you take notes, as this material will be useful for the homework.
(1) This is priority number 1. Read one of the research articles: part 1 or part 2 . Write a one-page review outlining (i) the research problem presented in the paper, the originally unsolved question, (ii) solution to the problem: the robotic application and its use, (iii) what we learned from this work, (iv) work involved to implement it, (v) materials/parts used, and (vi) your critique or suggested improvements on the work, or possibly ways to extend that research further. Use the idea bounce outline as a guide. This 1-page report will serve as an example to lay down the foundation for the design of your own project, but their content need not be the same. Thus, the goal in writing it, is to practice developing a simple general concept of the final product, and the "vision of the end". We will further discuss these reports as a group so as to get insight into problems we may encounter in our own projects, and put together a tentative list of materials and equipment we might need. Please email me your report so I could get it back to you a few days before discussion of Week 3.
(2) This is priority number 2. Start thinking about your own project. Check out tools and suggestions if you need help generating ideas.
(3) Read about function and models of neurons to prepare for Week 2 discussion.
August 25, 28 Lab: Introduction to the course. Overview of the design path for the robot: neurons, synapses -> CPG -> NMJ -> PIC -> robot body -> research experiments -> new knowledge gained. Tour of the lab and of the website, server space. Brief primer on breadboarding, RC-oscillators and IC-circuits for those who are not comfortable with electronics.
Discussion September 4: Computer models of a neuron: Integrate-and-Fire, Hodgkin-Huxley, oscillators, others. Hardware models of neurons: analog, digital. Quick oral idea bounce: we critique published projects you read as homework. We use results of this discussion to get insight into problems we may encounter in our own projects, and put together a tentative list of materials and equipment we might need. Check whether we have it all, and order parts we do not have. I get the paper reviews for grading of Week 3.
(1) Read about function and models of synapses to prepare for Week 3 discussion.
(2) Consider suggestions I have given you on the 1-page report from Week 1 and prepare for Week 3 oral idea bounce on your own projects (see further).
Lab September 1, 4: A brief primer on how to write a scientific paper. A set of placeholders to make your writing easier can be found here. Intro to basic electronics. Become familiar with the electronic components that will be used to build robots. Perform simple exercises to understand the action of buffers, operational amplifiers, inverters and RC-circuits.
September 11 Discussion: Biologically plausible models of a synapse. Quick oral idea bounce: we discuss the ideas you hve for your own projects and make suggestions that should help you in your homework.
(1) Write a 2-page (+/-- 0.5 page) project plan outlining the goals, timeline, and materials for your own work. This will serve as a foundation for your paper and will include a choice of an animal after which your robot is being modeled. Use the updated 1-page report from Week 1 as a basis for the materials and methods. The other page will be occupied with your motivation of the chosen animal and description of a research problem you would like to solve with your robot. Communicate with your partner about the preferred project, so that you would not pull each other in the opposite directions.
(2) Read about biological pattern generators to prepare for Week 4 discussion.
September 8, 11 Lab: we discuss your chosen projects and come up with improvements. Take notes. Coordinate with your lab partners, if any. If you notice differences in your project outlines, choose what's best. Build on a breadboard and manipulate artificial neurons using RC circuits. Choose biologically plausible values for resistors and capacitors to produce realistic time constants of firing. Build a synapse, connect two neurons together. Obtain a neuronal oscillator by adjusting the time constants of the neurons and synapses as well as the synaptic strength. If there is time left, start building the network based on known biological CPGs that corresponds well to your project. This will involve building enough neurons and synapses to connect the components in a manner analogous to the biological CPG, and adjusting the components (time constants and synaptic weights) to obtain a plausible firing pattern.
September 18 Discussion: Biological pattern generators for walking, crawling and swiming. I recommend you take notes, as this material will be useful for labs of Week 4 and 5 when you implement the CPG on the breadboard. I get the project plans for review.
Homework assignment: Read these resources about how to write scientific papers to prepare for Week 5 discussion.
September 15, 18 Lab: Continue building your CPG. and obtain a plausible firing pattern. If you have time and inclination, start working on the body.
September 25 Discussion: How to write a scientific paper - an extended discussion, based on feedback from the weekly write-ups to-date (there's been only 2). You get your project plans back with comments. Make notes on them during this discussion, so you could implement changes by next week. If we have time left, we also discuss robot body parts, based on your chosen project.
(1) Start filling the placeholders in your paper with material we have discussed in the previous weeks: models of neurons and synapses (but only ones applicable to your particular project), supporting literature that you will use further in intro and discussion.
(2) Scan PIC manual to prepare for Week 6 discussion. Use these guidelines to focus your reading. Do not worry if you do not understand some parts - they will be covered in Week 6 discussion.
September 22, 25 Lab: Continue building your CPG. and obtain a plausible firing pattern. If you have time and inclination, start working on the body.
October 2 Discussion: How to program a microcontroller. I get weekly paper-in-progress for review.
(1) Describe your CPG and possible designs of robot body in the paper. Alternatively, you can start thinking about how to write code for your PIC.
(2) Read about muscles and neuromuscular junctions to prepare for Week 7 discussion.
September 29, October 2 Lab: Build the body of the robot. This will be dependent on your project, so I cannot give you recommendations at the moment, but we will be discussing it.
October 9 Discussion: Muscles, muscle-like actuators, and neuromuscular junctions. I get weekly paper for review.
(1) Keep working on the paper, or the PIC code.
(2) Read about control and behavior to prepare for Week 8 discussion.
October 6, 9 Lab: Program a microcontroller to generate "spikes". Write a C code to program a microcontroller to detect the voltage on the "neuronal membranes" and generate "spikes" depending on the preset neuronal firing thresholds.
October 16 Discussion: Control and behavior, sensory integration. VLSI implementation of control. I get weekly paper for review.
(1) Just keep working on the paper, or the PIC code.
(2) Read about Neuromorphic and digital hybrid systems to prepare for Week 9 discussion.
October 13, 16 Lab: Continue to program the microcontroller to generate "spikes". Debug C code. Program PIC. Adjust parameters so they work properly with your CPG.
October 23 Discussion: Neuromorphic and digital hybrid systems. I get weekly paper for review.
(1) Keep working on the paper, or the PIC code.
(2)Read about gait to prepare for Week 10 discussion.
October 20, 23 Lab: Make neuromuscular junctions (integrators). Program a microcontroller to control servomotors. Continue making robot: at this point your progress will be very specific to your project, so I cannot tell you exactly what to do.
October 30 Discussion: A walking gait - what sets the pace? I get weekly paper for review.
(1) Keep working on the paper. Describe and make drawings of your robot body and CPG. Insert photos? printouts from oscilloscopes? Describe functioning of the CPG and how it is supposed to control the motors.
(2)Read about Sensory neuromorphic systems to prepare for Week 11 discussion.
October 27, 30 Lab: Connect the CPG, microcontroller and the robot body and tune them to produce a biologically plausible walking/crawling/swimming of a robot similar to the animal being modeled.
November 6 Discussion: Sensory neuromorphic systems. I get weekly paper for review.
Homework assignment: same; read about Learning in silicon networks
November 3, November 6 Lab: Attach sensors to the CPG and model phototaxis/photophobia or other behaviors. Start doing experiments necessary for your mini research project.
November 13 Discussion: Learning in silicon networks. I get weekly paper for review.
Homework assignment: same; read about Neural control of robotic devices
November 10, 13 Lab: Finish the project and fine-tune the robot. Gather the necessary data for the paper: photos of the robot and its CPG, oscilloscope view of neuronal firing etc. Continue doing experiments
November 20 Discussion: Neural control of robotic devices. I get weekly paper for review.
Homework assignment: same; read about Biomorphic engineering in medicine and Evolutionary approaches to neuromorphic design
November 17, 20 Lab: Scientific paper workshop: exchange ideas on the structure and content of the paper, develop a vision of the discussion section of the papers and work out the details of the literature review.
THANKSGIVING - NO CLASSES
December 4 Discussion: Biomorphic engineering in medicine. Evolutionary approaches to neuromorphic design. I get weekly paper for review.
Homework assignment: same
NO LAB: WRITE THE PAPER
December 11 Discussion: Instruction ended yesterday, no class
NO LAB: PAPER IS DUE