The Dirty Rat

In the Spring of 2003, I took ECE 4240 - Introduction to Mechatronics at Tennessee Tech University. It was a design class in which we were to build a cleaning robot(PDF) similar to the Roomba.

As with all projects, we started out very ambitious, and we planned all kinds of strategic options for the competition(countermeasures like launching trash over the wall, IR "bombs" to confuse their range finders, etc.). The course was based around using the Motorola 68HC12 Microcontroller. We were divided into teams of 4: 2 electrical engineers and 2 mechanical engineers.

We named ourselves the Dirty Rats for two reasons: a) We wanted to fight dirty if need be, b) we planned on experimenting with using a mouse for motion control. It is worth noting that we succeeded at neither goal.

In the course of the semester, we learned a lot about building robots, suffered from heavy class loads(which cut the amount of time we spent on the robot), and made it through the first round of the competition. However, during our second round(single elimination) we forgot to tighten a screw on one of the wheels and stopped driving, thus depriving us of a win, and putting us out of the tournament.

What follows is a discussion of the various phases of construction of our project, with illustrating pictures.

Table of Contents

The Body

We decided on a circular shape for our robot, driven by two wheels with a supporting caster. This had several advantages in control and competition. The round shape meant that it would be much more difficult to get stuck in corners, and the 2 wheels(aligned along a diameter) provided a zero degree turning radius which made driving considerably easier and turning very simple.

The interior of the robot was basically empty to serve as a collection bin for the material colleced during cleaning. All the control electronics and batteries were mounted on top of a sheet of plexiglass that covered the collection container. The motors and wheels were mounted inside the collection container as well.

This picture shows the completed base, with motors in direct drive of the wheels. Later on(this was planned) the wheels were replaced and a new drive system used.

Sensors

We had planned on using several sensors to control our robot:

In the end, we just used the IR range finder. We were at least smart about it though, unlike the other other teams, we built a calibration rig to convert the values given to us from the A/D conversion into actual units, so we could know how far away from things to stop the robot and turn it.

It is worth noting that we got very good calibration data, and it matched the spec sheet pretty well. The sensor gets unreliable if it gets closer than 3 inches. Our (good) solution to this, which we didn't do, was to mount the sensors 4 inches from the outer hull of the robot. Our (actual) other solution was to just plan on stopping far enough away from things that it wouldn't matter, and hope the motors were powerful enough to just move us out of the way if it didn't see it in time.

Driving

Getting Closer

The Competition

Afternotes

Team Dirty Rat, pictured from left to right: Ryan Crews, Wayne Douglas, Matt Estes(me), and Mike Mercer

We learned a lot from doing this. Good teamwork is important, and while we weren't always a good team members, we all contributed in the ways we could best. It wouldn't have been possible if it weren't for Ryan's late nite coding sessions in the lab, or Wayne's skills in cutting metal. Mike was an inspiration to us all with his enthusiasm, and I survived my other senior design project(a simple, but complete computer system, with its own RISC microprocessor designed by me).

As to the future. Wayne and I definitely intend to build more robots. Since then we've accumulated a lot of parts, electronics, microcontrollers, and design patterns, and general knowledge.

Design documents, links, references, etc.