The 2011 World Congress on Computer Science and Information Technology, WCSIT'11
Student Session Presentation
A HYBRID NEURAL NETWORKS-HIDDEN MARKOV MODEL STRATEGY FOR ACTIONS SEQUENCE AUTOMATION
Mostapha Abdeljawad*, Mohammad Adel Taher
Abstract
The present proposes a new strategy for action sequence automation based on artificial neural networks (ANNs) and hidden Markov models (HMMs). Action sequence automation is achieved through 2 consecutive modules. The first is a target generation module which uses an ANN to predict, given an image depicting the current state, the “should be” coming state images if the action sequence is proceeding in correct tracks. The second module is the Hidden Markov Model (HMM) action sequence generator, which given a sequence of states, produces the most probable related actions. To demonstrate the different aspects of the proposed strategy, we discuss its application to underwater welding automation
ليست هناك تعليقات:
إرسال تعليق