Reinforcement Learning for Robotic Reaching and Grasping
نویسنده
چکیده
A reinforcement learning approach is used to train a neural controller to perform a robotic reaching task. Unlike supervised learning techniques, where the teacher must provide the correct sequence of motor actions, only an evaluation of the robot's performance is provided. From this limited information, the robot must discover the appropriate motor programs that best satisfy the teacher's evaluation criterion. This type of learning approach is important because in a real-world environment, the teacher is generally not able to describe the motor program that performs the desired motor skill. This chapter utilizes the language of schema theory [1] as a mechanism for describing functional decompositions of motor programs. A connection is made from schema descriptions to a neural-level implementation of the schemas. It is at this low level of processing that we define a reinforcement learning algorithm that acquires motor programs that satisfy the reinforcement policy defined by the teacher.
منابع مشابه
Deep Reinforcement Learning for Vision-Based Robotic Grasping: A Simulated Comparative Evaluation of Off-Policy Methods
In this paper, we explore deep reinforcement learning algorithms for vision-based robotic grasping. Modelfree deep reinforcement learning (RL) has been successfully applied to a range of challenging environments, but the proliferation of algorithms makes it difficult to discern which particular approach would be best suited for a rich, diverse task like grasping. To answer this question, we pro...
متن کاملA Robotic System for Learning Visually-Driven Grasp Planning (Dissertation Proposal)
We use findings in machine learning, developmental psychology, and neurophysiology to guide a robotic learning system's level of representation both for actions and for percepts. Visually-driven grasping is chosen as the experimental task since it has general applicability and it has been extensively researched from several perspectives. An implementation of a robotic system with a gripper, com...
متن کاملBiological cumulative learning through intrinsic motivations A simulated robotic study on the development of visually-guided reaching
This work aims to model the ability of biological organisms to achieve cumulative learning, i.e. learning increasingly more complex skills on the basis of simpler ones. In particular, we studied how a simulated kinematic robotic system composed of an arm and an eye can learn the ability to reach for an object on the basis of the ability to systematically look at the object, which, in our set-up...
متن کاملBiological cumulative learning through intrinsic motivations : A simulated robotic study of the development of visually-guided reaching
This work aims to model the ability of biological organisms to achieve cumulative learning, i.e. to learn increasingly more complex skills on the basis of simpler ones. In particular, we studied how a simulated kinematic robotic system composed of an arm and an eye can learn the ability to reach for an object on the basis of the ability to systematically look at the object, which, in our set-up...
متن کاملTowards Vision-Based Deep Reinforcement Learning for Robotic Motion Control
This paper introduces a machine learning based system for controlling a robotic manipulator with visual perception only. The capability to autonomously learn robot controllers solely from raw-pixel images and without any prior knowledge of configuration is shown for the first time. We build upon the success of recent deep reinforcement learning and develop a system for learning target reaching ...
متن کامل