A novel inverse kinematics method for upper-limb exoskeleton under joint coordination constraints

In this study, we address the inverse kinematics problem for an upper-limb exoskeleton by presenting a novel method that guarantees the satisfaction of joint-space constraints, and solves closed-chain mechanisms in a serial robot configuration. Starting from the conventional differential kinematics method based on the inversion of the Jacobian matrix, we describe and test two improved algorithms based on the Projected-Gradient method, that take into account joint-space equality constraints. We use the Harmony exoskeleton as a platform to demonstrate the method. Specifically, we address the joint constraints that the robot maintains in order to match anatomical shoulder movement and the closed-chain mechanisms used for the robot's joint control. Results show good performances of the proposed algorithms, which are confirmed by the ability of the robot to follow the desired task-space trajectory while ensuring the fulfilment of joint-space constraints, with a maximum error of about 0.05 degrees.