Prediction of Seat-Off towards Assistance for Sit-to-Stand Movement by RBFN

Precise and decisive event prediction is necessary for benevolent movement with exoskeleton. Hence, two events are important during sit to stand assistance: seat off and stand off assistance. The seat-off event needs to be predicted to get the moment when start-of assistance should be started such that exoskeleton supports the Sit-to-Stand (STS) movement instead of inhibiting it. This paper proposed a Radial Basis Function Neural Network (RBFN) to learn a relation between inputs and the event. In this piece of work, we have predicted the seat-off event of multiple healthy human subjects through the RBFN network. Due to salient features of the RBFN network such as easy designing, generalization and strong tolerance to noise makes them suitable for this application. The method used allows reliable event prediction with joint angles as inputs. This model will be flexible as it will recognize the pattern for different people and work accordingly.