Design of autonomous robotic system for removal of porcupine crab spines

• Main Author: Wu, Haodong
• Format: Thesis
• Language: English
• Published: Memorial University of Newfounland 2023
• Subjects: Northwest Atlantic
• Online Access: https://research.library.mun.ca/15932/; https://research.library.mun.ca/15932/4/converted.pdf

Among various types of crabs, the porcupine crab is recognized as a highly potential crab meat resource near the off-shore northwest Atlantic ocean. However, their long, sharp spines make it difficult to be manually handled. Despite the fact that automation technology is widely employed in the commercial seafood processing industry, manual processing methods still dominate in today’s crab processing, which causes low production rates and high manufacturing costs. This thesis proposes a novel robot-based porcupine crab spine removal method. Based on the 2D image and 3D point cloud data captured by the Microsoft Azure Kinect 3D RGB-D camera, the crab’s 3D point cloud model can be reconstructed by using the proposed point cloud processing method. After that, the novel point cloud slicing method and the 2D image and 3D point cloud combination methods are proposed to generate the robot spine removal trajectory. The 3D model of the crab with the actual dimension, robot working cell, and endeffector are well established in Solidworks [1] and imported into the Robot Operating System (ROS) [2] simulation environment for methodology validation and design optimization. The simulation results show that both the point cloud slicing method and the 2D and 3D combination methods can generate a smooth and feasible trajectory. Moreover, compared with the point cloud slicing method, the 2D and 3D combination method is more precise and efficient, which has been validated in the real experiment environment. The automated experiment platform, featuring a 3D-printed end-effector and crab model, has been successfully set up. Results from the experiments indicate that the crab model can be accurately reconstructed, and the central line equations of each spine were calculated to generate a spine removal trajectory. Upon execution with a real robot arm, all spines were removed successfully. This thesis demonstrates the proposed method’s capability to achieve expected results and its potential for application in various manufacturing ...