Inline graphics on this page: 21K

Last update: 13 Nov 1999

Robot ``Easter Egg Hunt''

The American roboticist Mark Tilden - who coined BEAM (in 1990), which is now understood to mean Biology, Electronics, Aesthetics, Mechanics - thinks that it's a good idea to start a robotic evolution by building `wild' robots, machines that look after themselves and remain active for many years without needing their owner's helping hand. When the robots have learned to take care of themselves, then we may put them to work, domesticate them.

This interesting vision has so far led to generations of intriguing little robots which are fun to build and see move about, but which are still utterly unable to take care of themselves in the real world. Most of them `live' in controlled environments, where they are ornamental at best.
Several `bot-builders believe that the lack of progress is caused by a lack of honest work. When too little is demanded from robots, they don't need to evolve; they can just crawl about looking for a sunny spot, and then sit back and relax.

Playground

The ``Easter Egg Hunt'' is designed as the first of a series of tasks and environments to make small robots evolve a little faster. It provides a standard environment called the Playground, which includes the Eggs, a gathering area, a navigational aid and some obstacles:

Playground
Playground Playground table: matt white surface. Size: 100 x 50 centimeter.
Border Border: 2cm high, matt white finish.
Gathering area Gathering area: matt black surface. Paint or paper ensuring maximum contrast with the matt white Playground floor, to normal and IR-sensitive eyes. Always this size (radius 20cm) and in a corner. The border of this area is also painted matt black. For a contest, the gathering corner will thus be required to show contrast in IR light, as checked using Ken's method:
KH> A generic IR photo-transistor was used with a 22K collector
KH> load; the IR source, a 60W incandescent bulb about 50cm above
KH> the surface(s) to be tested. Using a 5V power supply, with the
KH> photo-transistor masked to permit light to enter only through
KH> the lens of the device, point it at the surface at about 60deg
KH> at 1 to 2cm above it. The collector voltage should read 3V or
KH> better while pointed at a low reflective surface and 0.2V or
KH> less with a high reflective surface and about 4.5V with the
KH> light turned off.
KH> It seems pretty rinky dinky but works well.
IR Beacon Modulated IR Beacon: Optional. Possibly something like this, with a few extra emitters to guarantee reception in a 90 degree arc.
Participants who wish to use an IR beacon must provide their own. The average current consumption must not exceed 50mA at 9V, and the emitter(s) must be at a height of no more than 5cm above the Playground floor, at the center of the gathering arc (as shown). If the beacon is to stand in the gathering area (instead of outside the Playground), it must occupy less than 25cm2. An unstabilized 9V source will be available.
Obstacles Ten obstacles: cylindrical, 2cm high, matt finish, kept in position with a central pin. The distance between them and between any obstacle and the border is 10cm or more. They may be anywhere on the Playground, except in the gathering area. Diameter between 2cm and 20cm.
Eggs Five ``Easter Eggs:'' Diameter between 1cm and 1.5cm, each made by crunching up a sheet of aluminum kitchen foil of about 15 x 15 centimeter. They won't resist rolling, but neither will they roll predictably. Conductive. Can be anywhere on the Playground, except in the gathering area.
Competitors who wish to do so may supply their own Eggs. Only aluminum kitchen foil is allowed as material, and they must conform to the above specifications. Ten Eggs should be supplied, so the judges can take a few apart for inspection, and also to have a few in reserve in case of attrition.
A couple of Eggs and a centimeter scale. Eggs

The Contest

The place, date and prices to be won are yet to be determined, but it will be organized along these lines:

Participating robots must fit inside a 9cm x 9cm square. (Tactile sensors are allowed some reach beyond these limits.) They have to find all Eggs and deliver them into the gathering area, within a maximum time yet to be determined. If no robot manages to find and gather all the Eggs, the one who got the most wins the contest. If more `bots complete the task, the one who did so fastest is the winner.

The Playground will be in an ordinary room, lit by ordinary natural or artificial light. Since the IR beacon is modulated, it should be easy to make sure that your Egg Hunter is not confused by mains modulated artificial light. But eye-challenged Egg Hunters may be offered a run in diffuse daylight on request.

Since BEAM is predominantly a world-wide Internet community, the contest will be organized as a `proxy' event. Participating robots must be mailed to the organizer. After arrival of a `bot - at least a week before the contest - the designer/builder will be contacted by the person selected to represent him/her, using email. The proxy will run appropriate tests, including a ball-gathering run on a standard Playground, and perform small repairs if necessary. After the contest, the robots will be returned in their original package (postage to be paid by the owner), together with any prices won.

Most robots will presumably require charging. They will have to bring their food with them, in the form of a charger which can take a stabilized DC input (participants may request any value up to 30V).
During the event, a camera directly connected to the Web will hopefully provide participants and public with a real-time view of the contest. The setting will be a page with pictures and details of the robots, stills of interesting scenes on the testing Playgrounds and so on.

Those working on Egg Hunters should have learned a lot about searching an area when their `bots have done their job. And the Egg Hunt is a reasonable first model for other problems, like vacuming a floor, mowing a lawn, and searching for mines. The first two applications are closer to most natural search problems. Every animal does plenty of `Egg Hunting' when it forages, looks for a place to build a nest, a mate and so on. Searching something is perhaps the most basic intellectual interaction between individual and environment.

The advantages of this approach were summarized nicely by Ben Hitchcock:

From: Benjamin Edward Hitchcock
Subject: Re: Contest (mailing robots in)
To: beam@corp.sgi.com (mailing list)
Date: Thu, 29 Apr 1999 15:41:29 +1000 (EST)
                   
I think the idea of a mail-in contest would be great. It
takes care of two factors:

1) reliability. I don't know about your robots, but mine
are usually pretty temperamental, and only work with a bit
of coaxing. It takes a pretty good engineer to make a robot
that will work first time, every time, after being shipped
halfway across the world! It would bring reliability to
being a very critical part of the design, instead of being
relegated to the back of your mind. If we are to design
robots that have half a chance of surviving in a small
portion of the real world, they have to be halfway reliable.

2) Fudgery. I'm not saying that BEAMers fudge their
results, but a number of them have experience with science
experiments, and as everyone knows the only way to get them
to work is to fudge them. I think most people only publish
their results that reflect favourably on the robot, and
'forget' to include the time their robot got stuck in that
curious 130 degree wedge near the kitchen sink that managed
to elude both their touch sensors....

I think this sort of competition would make people engineer
and refine their designs a lot more than any 'test yourself'
competition could ever hope to achieve.

And it would be fair.

Just my thoughts,
Ben

Status

At this time (June 16 '99) Four people are known to be developing Egg Hunters. The progress of two of them is reported on these pages:

Acknowledgements

Originally, the IR beacon had a fixed modulation frequency. The present definition (which only limits average power and emitter height) was written after Chiu suggested:

Date: Fri, 30 Apr 1999 14:41:46 -0700
From: Chiu-Yuan Fang 
CC: beam@corp.sgi.com
Subject: Easter egg hunt suggestions

My only suggestion:

Use an IR output of 40kHz so that Sharp GP1U58X modules
(easy to finy) or Sharp IS1U60 modules (very small) can be
used.  That would make things a whole lot easier.

Chiu

The charger input of any stabilized DC value as requested by participants (up to 30V) was defined after Bob Shannon suggested:

Date: Thu, 29 Apr 1999 19:08:58 -0700
From: Bob Shannon 
To: beam@corp.sgi.com
Subject: Re: Contest!!!!

---8<---
> A rechargeable `bot would obviously have to bring its
> `food' along.  Proxies can't be expected to have the right
> charger for any robot.

That was my concern.  Auto-switching 120v 60 hz to 230v 50
hz power supplies are much more expensive than the common
wall-wart devices.

May I suggest that all charging be done from a pre-set
standard DC voltage, say 12 volts or something?
---8<---

The offer to run an Egg Hunter in diffuse daylight was added to the contest description after Wilf Rigter remarked:

Date: Mon, 3 May 1999 18:20:16 -0700 
From: Wilf Rigter 
Cc: beam@corp.sgi.com
Subject: RE: Easter egg hunt suggestions

---8<---
there may be some difference in interference from 50H and 60Hz fluorescent
(TL) lighting for the beacon sensor as well as other  photodetectors so
perhaps DC halogen lamps or equal should be used for ambient lighting to
eliminate that (possible) problem.    

Also I would like to suggest a name for the behaviour of this class of
robots : Ovatropism
and that would make these:  Ovatropic robots. 
---8<---

To subscribe to the BEAM MAILING LIST send e-mail with `subscribe' in the body of the message to: beam-request

sbolt@xs4all.nl