Thursday, 1 July 2010

Compensation for machines...

When designing robots, or machines in general that try to emulate the tasks of humans, it is most often the tasks we find easiest that are most challenging to implement in the machine. Much of this is down to us not understanding the way that the human body works. But just as in the past when it was instead the problem of insufficient computing power, we are catching up fast. Often in the past, it was looking at how some aspect of the human body didn't work correctly that enabled deduction of how it normally does work. Now, in addition, scientists are increasingly able to look directly at how the body (and in particular the brain) is working and are able to analyse its normal working state.

Going a stage further, it is now not uncommon for studies to be done which look at how the body adapts to circumstances when the normal body gets damaged and has to compensate. Robots have been built for some time now which can walk, run and climb stairs. The latter was often a joke levelled at the Daleks who have been ever-popular in the Sci-Fi series Doctor Who. Robots that are agile in movement are most useful, firstly in robot warfare but later more commonly in other applications as the technology passes from military to civil applications in the usual way. Scientists are now already studying, with the help of amputee animals, how the body can compensate for the loss of a limb which is central to movement and thus how the animal can remain an effective mover. One such example is described in an article on BBC News.

In the future, machines will not only be more capable than the human brain in terms of computational ability, but also harness an understanding of strategies which can be employed when their control of a robot device is affected by damage or malfunction. This will put them in an excellent position to cope with situations which are not originally envisaged in their design. Robotic machines are still very much in their infancy, but are likely to grow up quickly.

No comments: