Learning inverse kinematics

Real-time control of the end-effector of a humanoid robot in external coordinates requires computationally efficient solutions of the inverse kinematics problem. In this context, this paper investigates inverse kinematics learning for resolved motion rate control (RMRC) employing an optimization criterion to resolve kinematic redundancies. Our learning approach is based on the key observations that learning an inverse of a nonuniquely invertible function can be accomplished by augmenting the input representation to the inverse model and by using a spatially localized learning approach. We apply this strategy to inverse kinematics learning and demonstrate how a recently developed statistical learning algorithm, locally weighted projection regression, allows efficient learning of inverse kinematic mappings in an incremental fashion even when input spaces become rather high dimensional. Our results are illustrated with a 30-DOF humanoid robot.

Authors

• Aaron D'Souza
• Sethu Vijayakumar
• Stefan Schaal

Keywords

• Humanoid robots
• Manipulators
• Spatial resolution
• Motion control
• Robot kinematics
• Constraint optimization
• Computer science
• Neuroscience
• Inverse problems
• Statistical learning
• Inverse Kinematics
• Degrees Of Freedom
• Learning Algorithms
• Inverse Problem
• Optimization Criteria
• Input Space
• End-effector
• Humanoid Robot
• Inverse Mapping
• Statistical Learning Algorithms
• Linear Model
• Training Data
• Cost Function
• Optimal Control
• Receptive Field
• Local Method
• Learning System
• Slow Movement
• Joint Space
• Null Space
• Cartesian Space
• Minutes Of Training
• Trajectories In Space
• Motor Commands
• Movement Planning
• Operational Space
• Robot Motion
• Valid Region
• Forgetting Factor
• Input Dimension