Visual servoing of a Gough-Stewart parallel robot without proprioceptive sensors

In this paper, it is shown that computer vision, used as a redundant metrology mean, allows the control of a Gough-Stewart parallel robot without using any joint sensor. This result is highly relevant for the control of parallel mechanisms hard to instrument (for instance, hydraulically actuated ones) or with large displacements (for instance, large telescopes positioning systems). It is also very interesting since it turns computer vision, usually considered as an exteroceptive sensor, into a proprioceptive one, yet preserving its exteroceptive nature and non-contact measurement.