Dexterity Online Optimization of Robotic Endoscopes Based on the Constrained Condition Number Under RCM Constraints

When assisting surgery, robotic endoscopes need to meet the remote center of motion (RCM) constraints, which limits the available joint space. To improve the dexterity of robotic endoscopes, we design a constrained condition number index and propose a dexterity optimization algorithm based on this index. The algorithm solves the optimization problem using a primal-dual neural network, eliminating the need to compute the pseudoinverse of the Jacobian matrix, which enhances robustness near singular configurations. In the validation section, we compare the constrained condition number index with other indices. Experimental results show that the optimization effect based on the constrained condition number outperforms other indices, effectively reducing tracking error, RCM error, and maximum joint velocity, and features a fast optimization speed.