Saturday, 15 November 2014


ok I think that i might of finally got this soft robotics thing – reacting to its context, structural memory in deformable objects – yesterday I brought in a structure to the meeting that I had already shown – but changing the parameters (making two elements longer) meant that I was able to manipulate it in one plane – in x and y axis – making it perform  in an exaggerated ‘flipping’ motion within a number of ‘other’ axis’s – this was an easy movement that I performed without thinking – I had not taken into account – body memory, variable stiffness or morphological computation - the model was then strapped to a xy table and initially a circle was coded into the movement of the arm – the uneven forces within the structure broke the arm which was attached to the structure – the structures construction meant that when the structure was manipulated it was revealed that the forces within it were distributed unevenly – something that was felt but overlooked and not fully understood when manipulated in the hand - then the movement of the arm was simplified into 4 points in space - a square, even though we had now worked out that the movement was octagonal – the 8 fitting onto the 4 sides each and with corners being taken up by an angle– due to the forces a second arm was broken – (robot war!!! my robots 2 theirs 0) – but with the addition of a spring the forces were subsumed and the structure moved beautifully – a case of passive actuation in practice.

The body makes and compensates for millions of movements continuously without recognising them – a robot which engages with its context has to undergo a similar process but it has to know – the maths involved in the coding is so mind boggling – I get it......actively controlling the passivity.

No comments:

Post a Comment