Artificial intelligent with artificial neuron method in doing image processing

Matlab Base Image Processing by Using Artificial Neuron Network for Robot Sorting System.

Dr Ahmad Radzlan Yusoff^1,a and Madan Varmma A/L Suparmaniam^1,b

a Clement Lai Tun Hao clementlai90@hotmail.com Bachelor of Mechatronic Engineering

¹Faculty of Manufacturing Engineering, University Malaysia Pahang, 26600 Pekan, Malaysia

ABSTRACT

In the globalization, robotic arm is widely used in the industry such as automotive manufacture industry, food industry, etc. Mostly, it perform the work in which beyond human strength and it can control manually or programmed to do a limited task with the help of sensors. In additional, the implementation of camera and filter by image processing to do data analysis and artificial intelligent neuron network to improve the computational efficiency. The machine will be taught to adapt the mechanisms that enable computer to learn from experience, learn by example and learn by analogy. The target of this method will be implement in rubbish sorting industries by robotic arm. The example data and picture files will store in the data base and divided into category. From the camera system, the picture will be captured and do analysis by artificial neuron network to find the images. In result, the output data will be more accurate and precise.

1.0  INTRODUCTION

In the present day, wastes or rubbishes are getting more and more. It also generate pollutions and other hazards to the global environment, besides that it also created a lot of waste lands for rubbish storing. By recycling of waste materials, it will save a lot of energy, resources and raw materials.

Overall waste composition in Malaysia is dominated by municipal solid waste (64%), followed by industrial waste (25%), commercial waste (8%) and 3% consists of construction wate (EU-SWMC, 2009). About 80% of municipal solid waste are recyclables, which are disposed at the landfills (MHLG, 2006) and under the category of municipal solid wastes, the contribution of household waste is the highest among sources consisting of recyclables at most 70 – 80% of total solid waste composition as found placed in the landfills (Sumiani et al., 2009). Household area is one of the main primary sources of municipal solid waste in Malaysia, besides institutional and commercial waste (Tariq and Mostafizur, 2007).

2.0 PROBLEM STATEMENT

            To increase the accuracy of the data output by implement an image processing to solve the problem of the human eye sign and man power issues. Besides that, to recycle the unwanted rubbish and promote green conception. By applying the artificial intelligent method on robotic system in doing rubbish sorting.

3.0 OBJECTIVE

1.      By using Matlab Artificial Neuron Network on the robotic system.

2.      Using image processing (OpenCV)  method integrate with Artificial Neuron Network.

3.      Create an Artificial Neuron Network on rubbish sorting robotic arm.

4.0 METHODOLOGY

            Using MATLAB software (R2010a) to generate the Artificial Neuron Network method. By the artificial neuron network toolbox, it can generate a regression graph, best validation performance graph and training state graph.

            Secondly, by using the OPENCV library in MATLAB, to analysis and generate the image processing methods. Implement one of the method in OPENCV, Speed Up Robust Features (SURF) to do the processing of image processing by pictures identification.

The summary report of my project

Computer Vision Detection and Controlling

Faculty of Mechtronic Engineering, University Malaysia Pahang, Malaysia

.

Project Supervisor: Dr. Ng Liang Shing

Degree Researcher: Clement Lai Tun Hao

Abstract

The innovation of technologies such as robot able to perform human tasks that being programmed to make up a better life for human. The computer vision simulates a data output for the respond command and allows the robot to detect the specified objects. The data output detected by the robot can be uploaded to the server and send to the users for further investigation. Image processing method is implemented to improve the vision of the robot. It is based on the similarity of colors, edges, and shape compared with the real time images. Users can monitor the robot from computer or phones and download and upload the pictures data from the Local host. For this research, computer vision is applied to acquiring, processing, analyzing and understanding the image of the detected object to simulate a data output for the respond command. The methodology of this project is firstly set the RGB and chooses the targeted picture before turning on robot; secondly the threshold method will be set up to identify and searching object distance; thirdly the coordination’s on pixel will make the movement decision toward the required object and pick the object.

Identification and detection of specified image of the objects by the designed robot will be more precise with the installation of camera and image processing.

Computer vision will able to solve the issues problem by human eyes limitation. Such as, traffic cctv, security camera, object tracking, etc. The police will able to track the wanted persons.

Furthermore, the robot can be tested in the mitigation strategies place such as radioactive area, earthquake, disaster, etc. The robot will able to detect human life form and let ease the job of the rescue’s teams.

The data of the robot (pictures and videos) can be uploaded to the server and send to another expert to investigate the problems.

Problem statement

To improve the vision of the robot by implement the image processing method base on the similarity of colors, edges, and shape compared with the real time images.

From the Local host network, users can monitor the robot from computer or phones. Besides that, the users can download and upload the pictures data from local host.

Methodology

1).Using Visual Studio C/C+ programming to test the OpenCV software.

2).Local Host by using PHP, HTML, CSS, and Java.

3).Arduino assemble language to program the Arduino Mega microcontroller chip.

Conclusion:

      By implement a camera for the robot and image processing, the robot will be more precise in identify the image and recognize the object. 

Literature Reviews:

1). Title: Self-Organizing Incremental Associative Memory-Based Rbot Navigation (10 October 2012).

Author:  Sirinart TANGRUAMSUB, A.L, M.T, & O.H.

Method: Input pattern as memory and clustered in the layer. The node hold pattern data.

S₁ = arg max || x – W_c ||

S₂ = arg max ||x – W_c ||

2). Title: Perceiving, learning, and exploiting object affordances for autonomous pile manipulation (26 September 2013).

Author: Dov Katz, A.V, M.K, J.A.B & A.S.

Method: Facet segmentation by computing depth discontinuities, estimating surface normal, and color-based image segmentation.

My Final Year Project Poster and video

https://www.youtube.com/watch?v=__DYe7RNV1U
the link above the about my final year project first trial video

Before i came to University of Malaysia Pahang

       When i still a high school level student, i always think that the engineering university will be a wonderful place, in which is full of knowledgeable persons around, and the high technologies equipment ( all kind of movie show as how great was an engineering school look like). Since i was a kid, i always like to be an inventor like Eisteen, Edison, and Michiyo Kaku, and my parents told me that you should study hard and aim for university.

        Finally i reach my first university, UTAR (university of TAR) at Setapak. UTAR consider as a half sponsorship by MAC and half private. I prefer study at Private University in which the school fees is much higher compared to Local University, cause in Malaysia, the chinese races were poisoned and taught by the society about the racism and quota. In detail, mean that, besides of Malays and Bumiputera, other races like chinese,and india are hard to be selected to study in local university. Even-though, u miserly been selected, you will not able to get the course of the study you opt for.

         In private university, the facility was very limited and out of choices. The machinery and equipment are poor and updated. But, the lecturers and tutors were great in term of their professional teaching skill, therefore, they all more focusing on theory and study. As an engineering student, i will prefer my study will able to link and integrate with products. I was quite disappointed that an engineering students had being closed in their innovative inventory abilities. Overall, 3 months study in UTAR was smooth. I am a questioning-student, who like to keep asking questions and to fulfill my curiosity. The lecturers and professors will treat me well, although i make them in troublesome (sometimes).

          3 month later, i receive an offer letter from University Malaysia Pahang (UMP). i was stunned, and an only thing came in my mind was, is it a fraud or a joke? Since my result was not well, and lastly i decided to go to UMP.