Improving Grasp Skills Using Schema Structured Learning
Abstract
In the control-based approach to robotics, complexbehavior is created by sequencing and combining control primitives.While it is desirable for the robot to autonomously learn thecorrect control sequence, searching through the large number ofpotential solutions can be time consuming. This paper constrainsthis search to variations of a generalized solution encoded ina framework known as an action schema. A new algorithm,schema structured learning, is proposed that repeatedly executesvariations of the generalized solution in search of instantiationsthat satisfy action schema objectives. This approach is testedin a grasping task where Dexter, the UMass humanoid robot,learns which reaching and grasping controllers maximize theprobability of grasp success.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2005
- Accession Number
- AD1126846
Entities
People
- Andrew H. Fagg
- Robert Platt
- Roderic A. Grupen
Organizations
- University of Massachusetts Amherst