Some Robotics Projects

Food handling (grasp site detection and fish-flinging)

My first real brush with robotics in its full hairy detail was on an EPSRC funded project (I was an RA) which addressed the problems of handling variable shaped frozen food products. The emphasis of the project was on intelligent algorithms for grasp-site detection. We were using a three fingered gripper, and were trying to develop a generic algorithm for this type of product. Basically the project was a success and we came up with an algorithm which worked quite well. The algorithm used a simple set of rules to move virtual fingers around the edge of the object to detect a good grasp site.

In order to test the grasp site, the robot then picked up the food item using the detected grasp site and followed a “shaking” routine to try to dislodge the object. The violence of shaking which succeeded in dislodging the object was then used as a measure of quality for the grasp.

Now maybe you can figure out the “fish-flinging” bit.

Cheap and cheerful mobiles (Stamps, servos and Lego)

I started messing about with mobile robots in about 1998, so I’m a relative newcomer. My first mobile robot was a converted radio-controlled car. I simply replaced the radio-receiver with a Stamp micro-controller board with a PIC based servo controller. I then added some whisker bump-sensors, infra-red scanning distance sensor and light sensors all around. All this stuff was “knocked-up-in-my-back-yard” type stuff. Americans please note that “knocked-up” doesn’t have those connotations over here. I learnt lots about dealing with noise, unreliable sensors and actuators and designing experiments with this beasty. We did some work on tele-operation with this robot and a Lego robot working together. The Lego robot was pretty crappy too. All a good learning experience.

My last foray into this type of robotics involved a “PC on wheels” that predictably enough got called “Robbie”. It had two scavenged stepper motors and old disc platters as wheels. Unsurprisingly it didn’t work terribly well and the battery life was atrocious. Never really got properly finished, but was my first real go at Linux stuff. Well worth the effort.

Fancy mobiles and a fancy lab

Thanks to a good dose of SRIF funding we now have a really good environment to do mobile robotics experiments in. We have a lab with VICON motion tracking gear and a set of four Pioneer 2DX mobile robots. We also have an outdoor Pioneer AT which I haven’t played with much yet. The 2DXs do all the stuff that I wanted to do with all the cheap stuff I mention above, but properly. They make the really interesting stuff a lot easier to get to without having to do all the tedious soldering first.

My current work is focussing on longevity and ways of making robots survive for long periods. Along with Jon Timmis (of AIS fame), I have come up with an interesting way of building robot controllers that uses neural networks, artificial endocrine systems and artificial immune systems. We’re still working on it, but watch this space. We have some preliminary stuff published and implemented and are still working hard to find time to get on with the idea.

A Sailing Robot?

As you may have noticed, sailing seems to permeate just about everything that I do. Unfortunately this now includes my robotics work. I have this idea to build a long-lived sailing robot for oceanographic monitoring. I have talked to lots of people about this now and about half of them laugh… the other half includes a company who build robots. They seem to think that the idea is a reasonable one and if we can find some funding we are going to build one and prove the other half wrong (I hope).

Our all-terrain robot looks like this, but also has bumpers on the front and back. It also has a 360 view camera mounted on top.

Our indoor robots look a bit like this, but not quite. They have bumpers too, and have a slightly different case. Both types of robot also have radio ethernet boxes on the top.