Agent-Based Convolution and Reinforcement Learning

Abstract

The problem with the current models like Darwin-OP or Boston Dynamic’s ATLAS is their up-time, especially with increased number of joints. These models try mimicking the human motion; they end up using a lot of actuators, which in turn leads to the use of a lot of battery power. This paper discusses the creation of a new model of humanoid robots, that does not try to mimic the bipedal walking gait used by humans, but who instead uses a full model constructed from scratch, that consists of a model free Deep Q-Learning (DQN) algorithm, which doesn’t need any walking sequence or walking models, it just learns from trial and error by applying actions on the robot and observing the reward from that action to make an under-actuated robot able to balance and walk forward, backwards, sideways, and rotate in place using only 4 actuators (two in each leg). The proposed model uses a Regional Convolutional Neural Network (R-CNN) to detect and inform the robot about the place of its goal. A full sensory system of a camera and Inertial Measurement Unit (IMU) is utilized to extract and gather the required inputs for reaching the goal from the robot’ environment. Thus instead of thinking that robot as a pre-programmed entity who performs specific task, we treat this as agent who can learn to take whatever actions towards specific goal controlled by evaluation function to maximize specific reward.

Authors and Affiliations

Samer I. Mohamed, Amr Abdelnabi

Keywords

Related Articles

The Relationship Between Financial Management and Organizational Behavior of Government Agencies Introduction by the Researchers

Financial management and organizational behavior, with the help of different types of organizations, focus on human factors and behaviors to achieve their goals and their survival, growth, development, and adaptation to...

The Role and the Influence of International Marketing on Export

Unlike the utilization of marketing within certain national frameworks where there are relatively uniformed conditions, in international marketing we encounter a series of specificities that result both from the specific...

Factors Affecting the Banking System's Deposits

What are important in early stage activities of banks is bank deposits. Bank deposits are impacted by organization outside and inside various factors. In this study, Friedman money demand models as a basis to evaluate th...

The Impact of the New Public Management on a System of Public Administration Reform: The Case of Ghana

This paper Examines the impact of the New Public Management on a system of Public Administration Reform. It looks at the rational for the New Public Administration including encouraging economic growth, pove...

Comparison of Yogic Practices in Hatha Yogic Literatures: Hatha Yoga Pradipika, Gheranda Samhita and Shiva Samhita

Hatha yoga pradipika, Gheranda Samhita and Shiva Samhita are three major classical treatises on yoga. Hatha Yoga Pradipika was written by Swami Svatmaram sometime in the 15th century C.E.Likewise Gheranda Samhita is a la...

Download PDF file
  • EP ID EP264556
  • DOI -
  • Views 130
  • Downloads 0

How To Cite

Samer I. Mohamed, Amr Abdelnabi (2017). Agent-Based Convolution and Reinforcement Learning. BEST : International Journal of Management, Information Technology and Engineering ( BEST : IJMITE ), 5(12), 17-28. https://europub.co.uk/articles/-A-264556