Artificial neural network based study of torque at knee during sit to stand and back to sit movements

Back-To-Sit and Sit-To-Stand movements are the most important and essential movements for any human being to sustain their life and to accomplish their habitual actions. Not only for human beings, but for robots or assistive devices to accomplish several tasks, these movements are vital. The knee joint has the major functionality to execute such movements and indeed, the torque required at these joints will be very high. Hence it is significant to study the torque at the knee joints. As inverse dynamics multiplies the computation step requirement based on the addition of number of degrees of freedom, the computation for the torque becomes quite complex for higher order systems. However, Artificial Neural Networks (ANN) reduces the computation load by building weighted networks between inputs and outputs based on the training with known data. This work is based on the computation of unknown torque using the back propagation algorithm built up from the known training input and output data. The predicted results with ANN are compared with the actual values and found to have a regression coefficient, R
²
, value of 99% and a sum of errors of 1%.