CSC290: Mobile Robotics

Class will divide into teams of 3 for projects. Choose partners with a variety of skills. It is important that your schedules are similar enough to allow significant hours of group work in the lab.

``Mathematics is proving theorems - and publishing them'' (Laplace). One of the goals I have is to produce a technical report similar to TR588, which describes the projects. I should also like a web page, i.e. a large set of pages, with graphics and images, for internal and external viewing. These goals require that a significant portion of the class know elementary LaTeX and HTML.

Class organization: I'm thinking we should have a quartermaster, through whom all requests and complaints and problems about hardware are funnelled. This may be the TA or could be a class member who gets paid by the hour. Job will involve driving to Radio Shack buying things, watching inventories, compiling requests, general care of the lab. This is a paying position, paid at the UG TA rate.

Also we need two editors, an HTML and a LaTeX editor, to take charge of TR and Web preparation. These are ``glory'' positions, in that the reward is lead editorship of the hardcopy (and ftp-able) and internet products, respectively. The editors' job is to collect the contributions from the groups, make sure all the included graphics work, resolve references, organize the content into a coherent whole. The TR editor will do most of his work after week 14. The WWW editor can spread work out over the semester.

Teams and grades. Each team has 3 members. Each team has 3 formal presentations, a project proposal, a progress report, and a final report. Each presentation will be both oral and written. Each component (the oral and written) will be the responsibility of one team member. Each team member must do one oral and one written presentation, but not necessarily for the same stage of the project. CB will assign a grade to the oral presentation, a grade to the written presentation, and a grade to the team as a whole. Call this T-GRADING.

During the semester there may well be assignments that are GRADED on an individual basis (like problem sets) and on a team basis.

The project has five phases.

Emerging from ignorance (Weeks 1-2). Get hold of all you can in the literature and on newsgroups (e.g. comp.robotics.misc, comp.robotics, comp.robotics.research). Learn what has been done and what it took to do it in the way of sensors, programming, computation power, etc. Learn all you can about the Rug Warrior, IC, the sensors and effectors available to us. UNGRADED.

Dreams and Sanity Checking: The Pre-proposal (Weeks 2-8). Decide what project you'd like to do, what you'd like to achieve or demonstrate. There will be some suggested projects but beware they are not time-tested or necessarily even doable. You must be able to justify and make credible anything you propose. Prepare a pre-proposal document for submission to instructor and TA. Get an OK, probably of a modification of your original plans. UNGRADED.

Formal Proposal (Week 9). Prepare and present the written and oral formal proposal. T-GRADED.

Formal Progress Report(Week 11) Prepare and present the written and oral progress report, possibly including a demonstration. T-GRADED. Further, I'm thinking that there may be mandatory floor exercises as part of the progress report. E.g. wall-following, responding to stimuli (e.g. obstacle avoidance, light sensitivity) going an estimated distance, some subsumptive multi-behavioral architecture, etc.. These exercises would be in common for everyone and would definitely require a demo.

Formal Final Report(Week 15) Prepare and present the written and oral final report, definitely including a demonstration. T-GRADED. Ideally we should have a formal event at which these projects are presented to the CS department at large. We'll see.

Collation of reports (Weeks 14-15). The editor should have a plan for the TR and should be bugging the teams to produce their final report on time and in acceptable style. Ideally the collation of the reports will not take more than a few hours, but CB, CE, and the editor may want to add glue, appendices, whatever.

Reruns. (Week 16) In case of unavoidable problems and unforseen disasters, we can re-run the final demos this week. Avoidable problems and forseen disasters don't count. These are RE-runs, not virgin voyages. This is not an extra week, this is not slack. Final work is due a week before this, unambiguously and absolutely.

Some Possible Project Components:

One ground rule is that if the 8' x 8' Robot Arena is used, it must be possible to set up for and take down from the task in a short amount of time. If competition for the space gets hot we may need to come up with a common configuration that will support all demos.

1.
Point-to-point navigation: like the Robot Fed-Ex of the video (go from point A to point B and back) but without the ``package delivery'' aspect. IR or line-following?
2.
Maze runner: blind (no learning) or with learning (like micro-mouse).

3.
Logo Turtle: I specify a path and the RW follows it. Might be neat to specify a path in an Xwindow and the RW goes and does it. Presumably one downloads the instructions and RW operates open loop.

4.
Following or goal-seeking: Light-sensitive or IR-sensitive or touch-sensitive following of a stimulus or other robot. Some sort of predator-prey behavior, simulation of competing drives (hunger, fear, etc.). Might need gradients painted on the floor...

5.
Straight line travel. Roll in a straight line, probably using visual or IR servoing off a distant beacon. Design and incorporate odometry to enhance other sensors.

6.
Extra components: mechanical, electronic, or sensors. Mechanical design and incorporation of more or different sensors or effectors could be a very effective project (e.g. the rotating sensor of the 1994 mobile robot).

7.
Last Year's Projects (on Carlson ). reserve.


---
This page is maintained by Deer Boy

Last update: 8.15.00