What are the computations of the cerebellum, the basal ganglia and the cerebral cortex?
Neural Networks - Special issue on organisation of computation in brain-like systems
Brains, Behavior and Robotics
Introduction to Reinforcement Learning
Introduction to Reinforcement Learning
TD Models of reward predictive responses in dopamine neurons
Neural Networks - Computational models of neuromodulation
Actor-critic models of the basal ganglia: new anatomical and computational perspectives
Neural Networks - Computational models of neuromodulation
Handbook of Learning and Approximate Dynamic Programming (IEEE Press Series on Computational Intelligence)
MOSAIC Model for Sensorimotor Learning and Control
Neural Computation
A fuzzy Actor-Critic reinforcement learning network
Information Sciences: an International Journal
Using continuous action spaces to solve discrete problems
IJCNN'09 Proceedings of the 2009 international joint conference on Neural Networks
Intelligent backstepping control for wheeled inverted pendulum
Expert Systems with Applications: An International Journal
Velocity and position control of a wheeled inverted pendulum by partial feedback linearization
IEEE Transactions on Robotics
Fuzzy inference system learning by reinforcement methods
IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and Reviews
IEEE Transactions on Neural Networks
Online learning control by association and reinforcement
IEEE Transactions on Neural Networks
| Hi-index | 0.01 |
In order to explore the possible cooperation mechanism between the cerebellum and basal ganglia in the central nervous system and to establish a more intelligent learning mechanism for robots, a new tropism-based ADHDP (action-dependent heuristic dynamic programming) learning mechanism involving the cortico-basal ganglia and cerebellar circuitry and the thalamic function is proposed. The cerebellum specializes in the actor part, while the basal ganglia are related to critic prediction. The thalamic function is considered as the tropism mechanism. Tropism value denoting the biological propensity is introduced to illustrate the degree of closing to the target. Although several motor control models have been proposed to explain the control and learning mechanism in the cerebellum and basal ganglia separately, it seems that the cooperation mechanism between them has not received much attention. In our proposed learning mechanism, the thalamic function and the cooperation between the cerebellum and basal ganglia are considered, and with a neurophysiological view, a striato-striatal lateral weight in the basal ganglia was added in the critic network. We present the detailed design architecture and explain how effective learning and optimization can be achieved with this novel tropism-based ADHDP architecture. Furthermore, we test its performance on the balance learning task of a two-wheeled self-balancing robot (TWSBR), which simulates the typical motor control and learning of the human body. In order to illustrate the effect of the thalamic function, some comparison researches about the balance learning problem have been done.