This paper deals with the design and implementation of avisual kinematic control scheme for a redundant manipulator.The inverse kinematic map for a redundant manipulatoris a one-to-many relation problem; i.e. for each Cartesianposition, multiple joint angle vectors are associated. Whenthis inverse kinematic relation is learnt using existinglearning schemes, a single inverse kinematic solution isachieved, although the manipulator is redundant. Thus anew redundancy preserving network based on the selforganizingmap (SOM) has been proposed to learn theone-to-many relation using sub-clustering in joint anglespace. The SOM network resolves redundancy using threecriteria, namely lazy arm movement, minimum angle normand minimum condition number of image Jacobian matrix.The proposed scheme is able to guide the manipulator endeffectortowards the desired target within 1-mm positioningaccuracy without exceeding physical joint angle limits. Anew concept of neighbourhood has been introduced toenable the manipulator to follow any continuous trajectory.The proposed scheme has been implemented on a sevendegree-of-freedom (7DOF) PowerCube robot manipulatorsuccessfully with visual position feedback only. Thepositioning accuracy of the redundant manipulator usingthe proposed scheme outperforms existing SOM-basedalgorithms
- Visual motor control
- Self-organizing map
- Redundancy resolution
- Inverse kinematics.
Kumar, S., Patcaikani, P., Dutta, A., & Behera, L. (2009). Visual motor control of a 7DOF redundant manipulator using redundancy preserving learning network. Robotica, 28, 795-810. https://doi.org/10.1017/S026357470999049X