An Integrated Connectionist Approach to Reinforcement Learning for Robotic Control
نویسندگان
چکیده
We explore the use of a connectionist-learning system designed to allow the application of reinforcement learning to robot control. In particular, we compare direct and indexed partitioning methods and nd indexed partitioning has advantages in time and space complexity, learning speed (measured in trials), and success rate. We make these comparisons based on extensive simulations and runs on a real robot learning on-line.
منابع مشابه
Autonomous Guidance and Control for an Underwater Robotic Vehicle
Underwater robots require adequate guidance and control to perform useful tasks. Visual information is important to these tasks and visual servo control is one method by which guidance can be obtained. To coordinate and control thrusters, complex models and control schemes can be replaced by a connectionist learning approach. Reinforcement learning uses a reward signal and much interaction with...
متن کاملLearning the Peg-into-Hole Assembly Operation with a Connectionist Reinforcement Technique
The paper presents a learning controller that is capable of increasing insertion speed during consecutive peg-into-hole operations, without increasing the contact force level. Our aim is to nd a better relationship between measured forces and the controlled velocity , without using a complicated (human generated) model. We followed a connectionist approach. Two learning phases are distinguished...
متن کاملHierarchical Functional Concepts for Knowledge Transfer among Reinforcement Learning Agents
This article introduces the notions of functional space and concept as a way of knowledge representation and abstraction for Reinforcement Learning agents. These definitions are used as a tool of knowledge transfer among agents. The agents are assumed to be heterogeneous; they have different state spaces but share a same dynamic, reward and action space. In other words, the agents are assumed t...
متن کاملSHRUTI-agent: a structured connectionist model of decision-making
A neurally plausible connectionist model of decisionmaking, based on the SHRUTI architecture, is being devloped. Toward this end, issues of appropriate connectionist representations for belief and utility, necessary control mechanisms, and reinforcement-based learning are addressed.
متن کاملControl Algorithm For Biped Walking Using Reinforcement Learning
The work is concerned with the integrated dynamic control of humanoid locomotion mechanisms based on the spatial dynamic model of humanoid mechanism. The control scheme was synthesized using the centralized model with proposed structure of dynamic controller that involves two feedback loops: position-velocity feedback of the robotic mechanism joints and reinforcement learning feedback around Ze...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2000