Image processing for multi-robot manipulator coordination

• Main Author: Prabu, P
• Other Authors: Hu, Guoqiang
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/140205
• Institution: Nanyang Technological University

This project aims to develop an image processing module (software) for robotic manipulators to detect and recognize objects present in the workspace in a coordinated fashion. The possible application of a tool sortation robot is considered for this project that can differentiate between tools and sort them accordingly. Images of the workspace would be obtained using cameras placed on the robotic manipulators. Multiple viewpoints of the tools in the workspace will be utilized to have a 3D understanding of the objects present in the workspace and consequently achieve an effective module for classification. These images would be processed to obtain signature characteristics of the objects in the image. The concept of Fourier Descriptor signatures is used in the recognition of the objects present in the image. These Fourier Descriptors of objects would be compared with the Fourier descriptors of the tool(objects) of interest. The deviation or the error between these descriptors would suggest the resemblance of the object of interest. The tools that is to be identified are trained to obtain the error threshold data which will be used as a form of decision guideline for recognition. Imagery data of multiple angles of the workspace obtained from cameras placed on multiple manipulators would be used in a coordinated fashion to achieve to effective tool classification results. Robotic arms will be designed in the ROS environment and simulated using Gazebo software. This designed robotic arm mimics the motion of a real-life tool sortation robot and serve as a visual representation of how a real life the tool sortation robotic may look like.