Machine Vision Control Algorithm Design for Industrial Palletizing Robo Machines to Increase the Dynamic Stability by Real-Time Image Processing

[thumbnail of TABISH_Muhammad-thesis_nosignature.pdf]
Preview
TABISH_Muhammad-thesis_nosignature.pdf - Submitted Version (6MB) | Preview

Tabish, Muhammad (2020) Machine Vision Control Algorithm Design for Industrial Palletizing Robo Machines to Increase the Dynamic Stability by Real-Time Image Processing. PhD thesis, Victoria University.

Abstract

The focus of the research is to palletise the laser cut irregular objects of metal, wood and marble. The large and heavy regular objects are very difficult to palletise by humans, even in the presence of manual palletisers. This becomes more complicated when the objects are of irregular shape. These objects are cut by precise laser into any shape, size and weight. Due to irregularity on the boundary and perforation inside the boundary makes it complicated for the Robots to palletise. Since palletising Robots are designed to grasp the objects from the fix spots and are preferred to be used for repeating jobs of same size and shape from same position. Therefore, Robot handling is also prohibited due to vast geometrical variation in objects. This issue has been raised in manufacturing industries that uses CNC (Computer Numeric Control) machines to mill or laser cut of large sheets. These sheets are commercialised in variety of most known materials like wood, marble and steel. Initially, all sheets are of regular shape mostly a rectangle with standard size of 1220mm x 2440mm observed in the wood industry. Since this configuration favours the Robot to palletise from pre-defined spot to the machine bed and cuts off the material into different shapes using precise laser. Once the laser cutting process is completed, the shape and size of the sheets are unpredicted, and this configuration is beyond Robot limitations therefore human handling is required. To develop a fully automated system and avoid heavy manual lifting, the Robot is necessary to collaborate with the environment by real time feedback system and integrate a controller to understand and solve the complex irregularity problems. This way the Robot can be used for non-repetitive task at unknown predefined spots. The Robot currently working on commercial scale uses the pneumatic grippers to palletise regular sheets. Some Robots have the capability to deal with irregular objects with limitations. These Robots pick the objects from COM (Centre of Mass) since they are very small in size and does not have sharp edges or perforation. The COM is a good technique for palletising only, if the objects are not too heavy or does not have much irregularities on the boundary. When a sheet is cut in a star shape with a hole at the centre or a grill type perforated having only 30 % of material after laser cutting, these scenarios are not yet been researched. The research proposes a MV (Machine Vision) controller that is designed, simulated on MATLAB (Matrix Laboratory) software and validated by implementing on a Robot in real time. The Algorithm is developed to work in a loop that repeat its cycle until or unless a human intervention subroutine is requested. The software takes images of irregular objects after fixed interval of time and evaluate the features of the shape. The image is disintegrated into finite small regular polygons through real-time image processing to formulate the trajectory. This trajectory is further analysed to configure the spot where the object can be grasped. Once the calculation is completed the MATLAB Algorithm communicate with the Robot controller and shares the positional information to the Robot. Now the Robot controller check the possibility to reach all the position and postures of manipulator. Further, this information is sent to check the range of end effector and enable it to start the operation. The whole system is in a feedback loop, if the object is dropped between operation due to miscalculation or mishandling. The Robot will stop and ask the MATLAB to re-evaluate the position of dropped object. If MATLAB is unable to calculate the trajectory for this object, the whole system will shut down and wait for human assistance. The robustness of the proposed method is evaluated through MATLAB simulations. To appreciate the validation of the Algorithm it is necessary to develop a prototype. Therefore, a 5-Axis Serial Robot (Mitsubishi RM-501) is used and the controller of this Robot is developed to read the information from MV system and integrate the pneumatic end effector with the Robot. F280049C Launchpad is used as main control unit of the Robot to control actuator, sensors and to communicate with MATLAB. MACH3 is also used as Robot interface software more details are available in chapter 5. The research has also been integrated in a project at R&D Department of a medical device manufacturer in Australia. The research internship provided development of an AUWS to weld soft polymeric materials together. The main objective after developing the machine from scratch was to weld these medical devices of different size and shapes. Therefore, MV technique is used to generate the different regular and irregular bonding pattern that could results in strong weld joints. Furthermore, the position and torque of ultrasonic welding head were also controlled based on thickness. The project is working and producing the medical devices for research purpose.

Item type Thesis (PhD thesis)
URI https://vuir.vu.edu.au/id/eprint/41808
Subjects Historical > FOR Classification > 0906 Electrical and Electronic Engineering
Current > Division/Research > College of Science and Engineering
Current > Division/Research > Institute for Sustainable Industries and Liveable Cities
Keywords robots; machine vision; MATLAB; algorithm; 5-Axis Serial Robot (Mitsubishi RM-501); irregular polygonal objects; real-time image processing
Download/View statistics View download statistics for this item

Search Google Scholar

Repository staff login