Accession Number : ADA264514


Title :   Drive-Reinforcement Learning System Applications


Descriptive Note : Final rept. 1 Jan 1989-1 Jul 1992


Corporate Author : MARTIN MARIETTA CORP BALTIMORE MD


Personal Author(s) : Johnson, Daniel W


Full Text : https://apps.dtic.mil/dtic/tr/fulltext/u2/a264514.pdf


Report Date : 31 Jul 1992


Pagination or Media Count : 76


Abstract : The application of Drive-Reinforcement (D-R) to the unsupervised learning of manipulator control functions was investigated. In particular, the ability of a D-R neuronal system to learn servo-level and trajectory-level controls for a robotic mechanism was assessed. Results indicate that D-R based systems can be successful at learning these functions in real-time with actual hardware. Moreover, since the control architectures are generic, the evidence suggests that D-R would be effective in control system applications outside the robotics arena.... Drive-Reinforcement Learning, Neural Network Controllers, Robotics, Manipulator Kinematics, Dynamics and Control.


Descriptors :   *ROBOTICS , *LEARNING , *MANIPULATORS , KINEMATICS , CONTROL SYSTEMS , NEURAL NETS , REAL TIME , DYNAMICS , TIME , DRIVES , ARCHITECTURE , TRAJECTORIES


Subject Categories : Psychology
      Cybernetics


Distribution Statement : APPROVED FOR PUBLIC RELEASE