An Improved Dual Quaternion-based Dynamic Movement Primitives Formulation for Obstacle Avoidance Kinematics in Human- Robot Collaboration System of Systems

Abstract

In the context of Human-Robot Collaboration (HRC) as System of Systems (SoS), motion planning and learning are a highly important subsystem. Dynamic Movement Primitives (DMP) are an elegant and efficient method for learning complex behaviours and representing them as stable, well understood dynamical systems. When applied for encoding a task by considering the behaviour of the end-effector of a robotic manipulator in Cartesian space, the common solution is to represent and to encode the pose separately as position and orientation using different but phase-coupled DMP formulations; resulting mathematically and algorithmically inefficient. Dual Quaternions are a mathematical tool capable of representing pose in a unified variable. Literature shows the interest of such a representation for rigid body kinematics, given its mathematical flexibility, efficiency and robustness. This article presents an existing formulation for Dual Quaternion Dynamic Movement Primitives based on screw theory. Then, we expand on our proposed improved formulation which considers obstacle avoidance as a contribution for the layer of Awareness and Intelligence in the context of Human-Robot Collaboration as a System of Systems.