The Technical Cognition in a Robotic System

1440 Words3 Pages

Robotics research has been conducted for the last few decades. The word ‘robot’ originally coined by Karel Čapek, a Czech writer in 1920 which means ‘forced labor’ (O2, 2013). The purpose of developing robots was for fulfilling the increasing demands of industrial revolution, but now are more focus on showing various level of cognition abilities and intelligence. And thus the research and development of cognitive architecture has been conducted over past few years. Generally, it comprises basic components for technical cognition and intelligence of the robot, including the ability to perceive, to learn, of motor control, of problem solving, of encoding and reasoning, and to communicate (Burghart et al., 2005). This proposal will focus on the ability to learn whereby it is possible to be acquired by a robotic system using Artificial Neural Networks (ANN), computational models proposed for the purpose of machine learning. There is a neural network model which is suitable for developing learning algorithm named Adaptive Resonance Theory (ART) that allows the learning occurs through adapting with the new knowledge without interfere the existing knowledge (Grossberg, 2013). Hence this proposal proposes a possibility of using ART to develop a prototype learning module based on cognitive architecture for an arm type robot simulator.
1.2 Problem Statement
Several cognitive architectures have been proposed to recruit the technical cognition in a robotic system. Upon this proposal, three types of cognitive architectures will be discussed. However, these cognitive architectures are still far beyond from the goal of creating actual human-like robots and thus there are many significant challenges are looking to be solved. One of the challengi...

... middle of paper ...

...cognise the letter (pattern recognition) and learn to write it using its own way. This prototype learning module will focus on letting the robot learns to write.
3.3 Evaluation on the Performance of the Prototype Learning Module
Several simulation experiments on pattern recognition and writing will be carried out on an arm type robot simulator, which is in a graphical environment in order to evaluate the performance of the prototype learning module on writing. The arm type robot simulator is going to be model and simulated using Robotics Toolbox for MATLAB that proposed by Peter Corke (2011). The robot simulator will learn how to write using the prototype learning module. The performance of the robot simulator on writing will be recorded for data analysis. And there is no user will be recruited in the evaluation on the performance of the prototype learning module.

Open Document