Learning dynamic models for robotic manipulation

• Main Author: Bui, Thien Phuc
• Other Authors: Domenico Campolo
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Engineering
• Online Access: https://hdl.handle.net/10356/177781
• Institution: Nanyang Technological University

Robotic manipulation is the backbone of robotics, which includes the control and coordination of robotic arms to perform different tasks, most notably assembly. Assembly process automation is significant due to its having widespread industrial applications. However, the complexity of assembly tasks presents challenges that are in need of researching. There are different methods that can be used, each of them has there own strengths and weaknesses, ranging from mathematical models to actual robotic arms with haptic feedback. Consequently, the necessity arises for virtual environment simulations to be studied, since prior research has not focused much on optimizing the part’s trajectory. This proves to be pivotal for precision, accuracy, and efficiency. This paper is set to explore an alternative, the Dynamical Movement Primitives - Blackbox Optimization (DMP-BBO) model, and to apply it to the peg-in-hole insertion task. The subsequent discussion of results would offer insights into the model, assess its effectiveness, and put forward suggestions for future work.