A research paper by scientists at the City University of Hong Kong proposed a hybrid model-based and online data-driven control method for a tendon-driven continuum robot, which requires no prior dataset collection or training. The method incorporates the Jacobian derived from the piecewise constant curvature model with online Jacobian error compensation using a Kalman filter.
The new research paper, published on Aug. 7 in the journal Cyborg and Bionic Systems, presented an approach eliminates the need for offline dataset collection and training. Experiments conducted on a planar continuum manipulator demonstrate the effectiveness of the proposed method in improving tracking accuracy for both position and attitude. Additionally, the effects of various model parameters are analyzed through comparative experiments.
