I've often been struck by how over-engineered industrial robot arms are. They are large, heavy machines which work to a remarkable precision and speed, and have an innate grace in their trajectories. This grace however, comes at a great financial cost. This is largely to do with the way in which the precision is achieved - all modern machines measure the joint angles with precision encoders, and then assume that the machine is perfectly stiff to compute where the tool is in space. This needs stiff linkages - which are heavy - and expensive gearboxes and motors.
For some engineering applications, like automotive, where badly placed spot welds could lead to failure, this level of precision is essential. However, this level of accuracy isn't needed for many applications. For the new crop of robot applications (like fruit picking) a few millimetres of accuracy is more than enough. Further, the position input will be from a machine vision system (rather than a pre-programmed pattern) which is 'accurate enough'.
So, we can start to strip out the things which make a traditional arm expensive, heavy and power hungry. If it's working on a farm or packing dog biscuits then a precision of a few millimetres and a payload of a few kilograms is sufficient. There are hobby robots available which can do either of these things separately, but not both together, and certainly not with acceptable speed. To this end we've just completed an internal project to make an industrial robot which can be built for less than £1,000. This has been achieved by using lightweight materials (stiff enough), simple DC motors (fast enough) and commodity encoders (accurate enough). The breakthroughs have come from simplifying the design to just the level required and using an advanced control system to maximise the accuracy attained.
In conjunction with the machine vision techniques that we've seen emerge over the last year, little can now be holding robots back taking on the jobs nobody wants – dull, repetitive, dirty jobs. The cheapness of this new technology will make it easier for managers with very thin margins to justify buying a machine, and thereby freeing their staff work at a much higher and rewarding level.