Bridging Reinforcement Learning and Iterative Learning Control: Autonomous Motion Learning for Unknown, Nonlinear Dynamics.

This work addresses the problem of reference tracking in autonomously learning robots with unknown, nonlinear dynamics. Existing solutions require model information or extensive parameter tuning, and have rarely been validated in real-world experiments. We propose a learning control scheme that learns to approximate the unknown dynamics by a Gaussian Process (GP), which is used to optimize and apply a feedforward control input on each trial. Unlike existing approaches, the proposed method neither requires knowledge of the system states and their dynamics nor knowledge of an effective feedback control structure. All algorithm parameters are chosen automatically, i.e. the learning method works plug and play. The proposed method is validated in extensive simulations and real-world experiments. In contrast to most existing work, we study learning dynamics for more than one motion task as well as the robustness of performance across a large range of learning parameters. The method's plug and play applicability is demonstrated by experiments with a balancing robot, in which the proposed method rapidly learns to track the desired output. Due to its model-agnostic and plug and play properties, the proposed method is expected to have high potential for application to a large class of reference tracking problems in systems with unknown, nonlinear dynamics.

Magnetometer-free Realtime Inertial Motion Tracking by Exploitation of Kinematic Constraints in 2-DoF Joints.

Inertial Measurement Units (IMUs) are used to track the motion of kinematic chains in a wide variety of robotic and biomedical applications. However, inertial motion tracking is severely limited by the fact that magnetic fields are inhomogeneous in indoor environments and near electronic devices. Methods that use only accelerations and angular rates for orientation estimation yield no absolute heading information and suffer from heading drift. To overcome this limitation, we propose a novel method that exploits an orientation-based kinematic constraint in joints with two degrees of freedom (DoF), such as cardan joints, saddle joints, the human wrists, elbow or ankles. The method determines the relative heading of the joint segments in real time by minimization of a nonlinear cost function. A filter for singularity treatment ensures accurate tracking during motion phases for which the cost function minimum is ambiguous. We experimentally validate the method in metacarpophalangeal (MCP) joints between the palm and the fingers. Accurate relative orientation tracking is achieved continuously despite several singular motion phases and even though the heading components of the 6D orientations drift by more than 360 degrees within ten minutes. The proposed method overcomes a major limitation of inertial motion tracking and thereby facilitates the use of this technology in robotic and biomechanical applications.