Learning perceptual coupling for motor primitives

Dynamic system-based motor primitives [1] have enabled robots to learn complex tasks ranging from Tennis-swings to locomotion. However, to date there have been only few extensions which have incorporated perceptual coupling to variables of external focus, and, furthermore, these modifications have relied upon handcrafted solutions. Humans learn how to couple their movement primitives with external variables. Clearly, such a solution is needed in robotics. In this paper, we propose an augmented version of the dynamic systems motor primitives which incorporates perceptual coupling to an external variable. The resulting perceptually driven motor primitives include the previous primitives as a special case and can inherit some of their interesting properties. We show that these motor primitives can perform complex tasks such a Ball-in-a-Cup or Kendama task even with large variances in the initial conditions where a skilled human player would be challenged. For doing so, we initialize the motor primitives in the traditional way by imitation learning without perceptual coupling. Subsequently, we improve the motor primitives using a novel reinforcement learning method which is particularly well-suited for motor primitives.

Authors

• Jens Kober
• Betty J. Mohler
• Jan Peters 0001

Keywords

• Couplings
• Acceleration
• Robots
• Trajectory
• Approximation algorithms
• Humans
• Learning
• Coupling Of Perception
• Motor Primitives
• System Dynamics
• External Variables
• Human Learning
• Imitation Learning
• Focal Variables
• External Focus
• System State
• Time Constant
• System Of Equations
• Motor Control
• Parametrized
• Learning Problem
• Function Approximation
• Motor Learning
• Transformation Function
• Robotic Arm
• Policy Learning
• Reinforcement Learning Problem
• Ball Movement
• Canonical System
• Reinforcement Learning Algorithm
• Discrete Movements
• Manual Tuning
• Model-free Methods
• Operational Space
• Control Problem