Numerical Optimizations in Inverse Kinematics of Robotics Hand

The analysis of hand movement focuses on two primary optimization issues: the calibration of the hand’s kinematic model parameters and the calculation of joint angles for specific postures. Typically, these problems are addressed independently. Calibration or scaling is resolved using static postures, whereas angle calculation can be performed through geometric or numerical methods. This work proposes a comparative analysis of three numerical optimization algorithms to solve these two problems simultaneously, without the aid of specific static tests but using dynamic trials only. Four tasks were selected based on specific literature to cover relevant grasping activities. The analysis was conducted through error analysis, using the discrepancies between the measured and estimated positions of markers. Results demonstrated error reduction in marker position when employing appropriate algorithm.