Copyright Ian Pearson, BT Futurologist

Click here for contact details,other articles and personal details

Smart Toys

Ian Pearson, BT Futurologist, Nov 2004

The toy world today is quite neatly fragmented. We have computer-based toys, that have some basic computing functions, lesser electronic toys, which have some simpler electronic functions such as simple sounds and flashing lights, and lots of non-electronic toys. Then there are computer games, and their various interface peripherals such as guns, joysticks and steering wheels. The expense of putting a lot of encapsulated intelligence into toys has meant that AIBO type robots and other sophisticated devices are really not for children, so can’t really be classified in the same groups, being only available to rich adults. Of course there is no ceiling once we include that group, since we then have cars, snowmobiles, personal jets and yachts.

But there is really no reason why even children’s toys shouldn’t be smart. Today it is possible to get a web server on a chip for just a few pounds, so it is perfectly possible to interact with toys via the internet. So why not have dolls that can be linked to the net, and have their own social lives? A few girls could coordinate a virtual soap opera involving their dolls. These could also be linked into games technology too, so that the kids could design their personalities on The Sims for example, and have the real doll, but also the virtual doll on the computer too.

This linkage between a little intelligence in the toy itself, to a lot more on the web or PC, seems to have been largely passed over by the toy manufacturers so far. There have been a few feeble attempts, but not many. Zowie’s ‘Ellie’s enchanted garden’ is among the best so far and even this was withdrawn. It used a computer game that was linked to the positions and orientations of plastic characters in the plastic garden. It used simple coils in the base of the pieces to detect what and where they were. A good start, imaginatively demonstrating how much can be done with just simple identifiers and a local positioning system.

It would be much more fun for kids if they could actually talk to their toys and get a sensible response, and similarly if they could get appropriate behaviours from them when they move them around, bring them together with other toys and so on. Chat bots are perfectly capable of simple conversation already; voice recognition and synthesis are pretty good too, so it just needs these to be combined with a speaker, microphone and radio link to enable this functionality via the web or PC. And the child could converse about the doll’s life in the computer game too. Detecting whether the toys are moving around is technically trickier, but can certainly be done. There are lots of ways of providing an indoor positioning system, either using radio or optical beacons for triangulation, smart mats, even ultrasound. Dead reckoning systems can also be made cheaply enough to add to toys. Tags and detectors on their own would be enough to enable smart interaction between toys, or parts of toys. Even magnets are sometimes used today for primitive detection purposes.

If a computer is able to link by radio to many of the toys in the playroom, they can effectively be made aware of each other, even though all the interaction actually happens on the computer. In fact, it wouldn’t cost very much to add some self-organising technology to the toys directly so that they could set up networks between themselves. Allowing toys to interact sensibly would increase the range of play options dramatically. A boy might be able to arrange battles between the toy soldiers, with their tanks and missiles, which could appear to behave and interact almost as if they were real. The scope of this could be greatly enhanced by using the power of nearby computers, but a fair amount of AI could economically be put in the toys themselves. Not all of the toys in a playroom would be designed to interwork, especially if they have come from competing manufacturers. But even adding an RFID tag to an object can give some effective intelligence via the computer, and since the tag could be as small as a few millimetres across, it wouldn’t need to spoil the appearance of the toy. It could take an interactive role in play without any intelligence of its own, provided that the computer knows where it is and how it is meant to relate to the other toys.

So smart toys are entirely feasible in the short term, at reasonably low prices. Companies like Lego, who have allowed thousands of kids to make their own simple robots, are likely to be in the forefront of the new toy generations, but there will be a lot of competition. All the base technologies are maturing and falling in cost, we can expect an explosion of functionality in the toy room soon.