Computer Vision Assisted Object Segregator Robot Engr. Ahson Ahmed11Department of Electronics & Biomedical Engineering, IICT, Mehran University of Engineering and Technology, PakistanCell: +92 345 3587885Email: [email protected]  Abstract— This paper discusses the project “Computer Vision Assisted Object Segregator Robot” which is related to the field of Mechatronics, Image Processing and Industrial Automation and it aims to detect three different shapes of the object merely through 2D low resolution camera with machine vision technique and uses two axis robotic arm manipulator to segregate and place these objects according to the designated positions of their respective shapes.

The system uses a computer with Matlab to detect the shape of the object via interfaced camera and uses microcontroller based electromechanical two-axis robotic Arm manipulator to grip hold and place object in a designated location. This system prototype and Software is made locally and none of the parts has been imported all the material and component which are used for designing this prototype are obtained from scrap and available in Pakistan. This project is very useful because most of the modern industries and Industrial research organizations around the world are now focusing on utilizing vision-based data acquisition systems with an intelligent manipulator for their various automation processes & quality control and segregation operations.

However, such solutions are quite expensive in Pakistan due to their initial and maintenance cost and import of machines from abroad. Likewise in Pakistan industries have been utilizing sensor based information for their automation and data acquisition process. However, said system has some drawbacks due to limited information sensing. Furthermore, in some complex tasks like inspecting products or segregation of different type of products, mostly human based or sometimes sensor-based approaches are used.

Since said approaches have found to be costly in term of labor cost and also less efficient in terms of limited information sensing and quality checks. Therefore a locally developed cost-effective and flexible vision-based intelligent robotic solution would be highly beneficial for local industries within Pakistan to stream line their various automation processes which might provide low cost of production with much higher quality control. Secondly Pakistan is an Agricultural state and exporter of  major crops of fruits and vegetable and grains this project has tremendous application scope in quality inspection grading and sorting of fruits and vegetables according to standards for local as well as international Markets and will revolutionize and assists local farmers to supply the best product for the local and international market a part from this it has application for local automobile/parts manufacturing industries pharmaceutical ware house storing etc. it has  future in autonomous vehicle surveillance monitoring, defense Applications, Artificial intelligence, machine learning and machine vision applications. Keywords- Computer vision system, Mechatronic systems,Quality Control, Inspection and industrial automation.                                                                                                     I.                         IntroductionIn Pakistan, most of the industries are utilizing obsolete manipulator and humans for various autonomous tasks in their operations because initial and maintenance costs of intelligent robots are very high because these products and their maintenance services both are acquired from abroad.

In view of above brief scenarios, it is aimed that a low-cost machine vision based intelligent robotic system may be developed locally to address the needs of Pakistani industries efficiently.·        The aims and objectives of this research work include:·        Recognition/detection of different geometric arbitrary  objects via machine vision·        Extraction of shape from image through boundary tracing & edge detection·        Designing a low cost flexible/modifiable two-axis robotic arm/manipulator·        Development of decision-making simple modifiable algorithm  for different shapes objects ·        Automation process for object placement at particular containerComputer vision devices for acquiring optical signals from a video camera have great importance in robotics which increases sensory abilities, usefulness, and flexibility of a robot system. Vision systems mostly use pattern recognition and have significance uses in many applications from automatic stock control robots to quality control check in automated manufacturing systems. It is a technology of machines that see.

As an engineering field, computer vision theory is related to the artificial systems that extract data, information, and features from images. This image data can be in many forms such as frames from video vision from Stereo or parallel cameras, or dimensional data from a medical scanner and so on.                                 II.        objective of the  proposed systemThe proposed system is able to detect the object along its shape orientation with a camera via machine vision technique. In this project, digital image filters boundary tracing technique and metric ratio based decision-making algorithm has been utilized for object Shape detection. Furthermore, a 2 axis robotic manipulator/arm comprising of automotive DC geared motors and motion limitation sensors is developed for grabbing an object after detection and for placing the same at the designated location. The prototype is capable of detecting and recognizing three different objects and its proper placement at specified locations. However, the said capability can also be extended to various types of objects.

Moreover, following are some prominent objectives that are covered in the subject project: §  Machine vision approach via Matlab§  Utilization of filters for feature extractions §  Designing of low cost flexible robotic arm/manipulator with embedded system§  serial data communication with microcontroller via PC§  Data acquisition by PC and microcontroller §  Designing HMI or Graphic user interface.§  Examining the flexibility of the System§  Proposing/development of less complex and low-cost design The most feasible and cost-effective method was the usage of spare parts of motorcycle gearbox to design prototype mechanical system and low-cost micro processing/micro controlling unit for project designing.                                                                          III.

                                 Mechatronic Manipulator                          Figure .1 Manipulator DesignThe 2 Axis robot arm design is integrated mechatronic embedded system which consists of a camera for computer vision, Clockwise moving base, arm (manipulator) and gripper (end effector). beside motors, gears, and relays robotic arm is integrated with positioned infrared base sensors and detectors for six position stop moments four for 320 degrees freedom of movement for base motion and two for up down 70 to 35 degree freedom of movement which reduces the cost of the system and excluded the need of expensive Servo motors and made the possibilities of using Automotive High torque position locking DC Geared Motors for high performance and Weight carrying capabilities Figure. 1 Prototype Manipulator                              IV.      Shape Detection AlgorithimMatlab Image processing toolbox was used to design the whole algorithm for the project. 640x 480 Resolution Image snapshot was used to Estimate each object’s area and perimeter by using shape factor formula which is commonly used in microbiology to determine shapes of particles .this results to form a metric ratio formula indicating the shape of an object by comparing it with threshold values and returning gradient value from the result of image matrices.

The metric will be equal to one for any single shape and it is less than one for any other shape. The Segregating process is controlled by setting an appropriate threshold. So that shape will be classified and the robot will take decision accordingly to place it to a specified position for a particular shape.

I have used three metric formulas for three different shapes as a prototype display but it can also be extended for arbitrary and different geometrical shape of the objects threshold values without applying particular metric formula is shown below in the diagrams.                             Figure .2 Test images of objects                        Figure .3 Threshold Images of Object                         V.            IMAGE PROCESSING FLOW CHART The Robot will take a snap shot from a live video sequence of object via camera .the Image will be processed by pc for shape detection and respective 8-bit code will be transferred serially for placing the object at the designated position.Morphology is used in image processing to perform operations on images morphological operator applies a structuring element to an input image and creates an output image similar to the input size.

In a morphological operation, the value of each pixel present in the output image is based on a value of comparison of the corresponding pixel in the input image connected with the neighboring pixel.The Sample of Output images of subject objects after applying morphology and edge detection filters are shown below in the diagrams      Figure .4 Morphological output images of objects                              VI.      Decision making algorithim   Shape factor formula Eq=4?area/perimeter2 for the detection of Round object and Eq=16*Area/perimeter2 for the detection of Square object Eq = 36*area/(3^0.

5)*perimeter2for the detection of Triangle object and got successful result under that condition and verified the metric is working for detecting and got successful result under that condition and verified the metric is working for detecting  shapes of the objects.  Figure .5 Output Images                               VII.     matlab gui design We have used GUIDE (GUI development environment) layout editor to create our control GUI of Robotic System for demonstration purpose it provides tools for programming and designing GUIs and also automatically generates the Matlab codes which defines all the properties of components and interact with the algorithm codes to establish framework call backs routine which is executed when a user interacts with the designed GUI.Figure .6 Main GUI Diagram of ProjectIn Main GUI Following tabs of operations and modes are presented which are used to demonstrate the working methodology of Prototype.                         VIII.       PROJECT HARDWAREAtmel 89C51 microcontroller based handcrafted embedded board is developed which is consist of input-output ports and serial data port with relay driving integrated circuit.

Following components are used and connected in the designs which are listed below.BOARD COMPONENTS•        AT89C51 Microcontroller I.C•        MAX 232 I.C•        ULN2803 I.C•        RELAYS      •        ILD74 I.

C•        I.R SENSORS •        POWER SUPPLYThe embedded board of prototype is mainly based on the single at89c51 microcontroller which drives and controls motion via limiting position of motors with I.R feedback sensors which are positioned with slit mechanism for limiting the motion of arm manipulator. Microcontroller receives 8 bit binary code from serial com port of computer which is interfaced with MAX 232 serial com leveler I.C with microcontroller when code is received respective action is produce by manipulator With I.

R feedback control.  TABLE 1. MICROCONTROLLER PORT CONNECTIVITY            S.N MCU I.C CONNECTED COMPONENT 1. Port2 pin 0-3 ULN2803 Vertical Axis Motor 2.

Port 2 Pin 4-7 ULN2803 Horizontal Axis motor 3. Port 3 Pin 5 ILD74 DC Motor 4. Port 3 Pin6 MAX 232 Serial port 5.

Port0 pin 0-5 FEEDBACK I.R DIODES I.R Sensors The assembly language programming is used for AT 89C51 microcontroller. The 8 bit position data is transmitted from the computer for each shape after detection and compared it with assigned 8-bit code within the controller memory. When the digital code is matched then particular data is sent from MCU Ports Pin to interfaced I.

CS as shown in the Table 1.and data which is transmitted to interface port with uln2803 is shown in Table 2 Shape Detection Transmitted  Data Manipulator Moment Square 00100000 1st Position Circle 00010000 2nd Position Triangle 00001000 3rd Position TABLE 2. TRANSMITTED DATA CODE The serial feedback position data through I.R feedback sensor is received by the MATLAB software via MCU for validating the position and task fulfillment of the manipulator in case of failure or fault the computer will resend the data again with three attempts if it fails to respond system will be automatically move on to halt state and indicate fault or failure.    Figure .

7 Simulated Diagram of Prototype Figure .8 Conceptual diagram of the Project  Figure .9 Original Picture of the Prototype                           IX.        SOFTWARE DESIGNThe software and programming language which are used while designing this project are as follows:§  KEIL µVISION§  GOOGLE SKETCHUP§  MATLAB 7§  EXPRESS PCB§  ASSEMBLY LANUGAGE                                            X.             ConclusionThe constructive achievement of this project is that the target for which it was designed to implement the concept application with low cost and resources was achieved .The detection of the symmetrical and arbitrary shape of an object allows the system to be used for various application and also can be upgraded with experiments and research and can be used for obtaining more features in machine vision applications locally. There are some limitations in the project which can be improved when it is designed for commercial use. The system successfully performed detection in various light condition and many arbitrary shaped objects resembling circular triangular and square properties via image was detected and robotic arm manipulator successfully done its placement task.

The plus point of this project which makes it simple and efficient was the used of I.R sensor and slit mechanism which avoided the need of expensive servo motors and reduces the cost of the system and made it adjustable for the specified movement. The use of locally available scarped material also encouraged as for recycling of spare parts and capable to be used in control applications Thus by preventing the system from complexities the system can be easily be commissioned and manufacturing cost dramatically reduced.The complete hardware and software has been tested and the functioning system is verified. The system can be extended with a different design for wide range of other applications. Different ideas from this application can be used to develop more innovative systems. Demonstration and the working link is given below.http://youtube.

com/watch?v=RBaoP9t8F5U=youtu.beAcknowledgment Author greatly acknowledges the support and co-operation of the Supervisor and also pays thanks to the Department of Electronics & Bio Medical Engineering, Mehran University of Engineering & Technology Pakistan.  References/biblography1     Fahad Ahmed, Haider Ali Zaidi, Syed Waqar Hussain Rizvi, Atiya Baqai and Bhawani S. Chowdhry “Electronic automatic gear shifting system for motor cycle ” Communications in Computer and Information Science, 1, Volume 20,  Wireless Networks, Information Processing and Systems, Pages 3-10,20092     Arsalan, M.; Aziz, A., “Low-cost Machine Vision System for dimension measurement of fast moving conveyor products,” Open Source Systems and Technologies (ICOSST), 2012 International Conference, page., no,22,27, 20-22 Dec.

20123     Sokic, E.; Ahic-Djokic, M., “Simple Computer Vision System for Chess Playing Robot Manipulator as a Project-based Learning Example,” Signal Processing and Information Technology, 2008. ISSPIT 2008. IEEE International Symposium  , vol., no., pp.75,79, 16-19 Dec.4     Cocota, J.A.N.; Fujita, H.S.; da Silva, I.J., “A low-cost robot manipulator for education,” Technologies Applied to Electronics Teaching (TAEE), 2012 , vol., no., pp.164,169, 13-15 June 2012