Date of Award

10-5-2023

Publication Type

Thesis

Degree Name

M.A.Sc.

Department

Electrical and Computer Engineering

Supervisor

Shahpour Alirezaee

Supervisor

Majid Ahmadi

Rights

info:eu-repo/semantics/embargoedAccess

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Abstract

This thesis addresses the growing demand for energy-efficient and high-speed industrial robotics. It focuses on optimizing motion parameters, specifically the velocity and acceleration profiles of a robot's tool center point (TCP). Unlike traditional methods that alter trajectory geometry, this approach preserves the trajectory while fine-tuning parameters for energy efficiency and reduced cycle time. It comprehensively explores energy consumption and cycle time optimization, using a robot inverse kinematic model for simulation. This research offers insights into the influential factors, including robot type, task performance, and operating conditions, contributing to a more efficient and environmentally conscious industrial robotics paradigm. We utilized multiple optimization methods, including Particle Swarm Optimization (PSO), grid search, scalarization, and Enhanced Multi-Objective Particle Swarm Optimization (EMOPSO), each tailored to explore the complex solution space of robotic applications. Our validation process extends beyond simulations to real-world experiments, affirming the feasibility and real-world applicability of our approach. One key insight gained is the impracticality of grid search for complex robotic systems, owing to its time-intensive nature. Furthermore, we observed the limitations of single-objective optimization, particularly in cases where energy consumption and cycle time must be balanced. As a response, we implemented multi-objective optimization algorithms, with EMOPSO emerging as an instrumental choice due to its adaptability, impressive convergence properties, and capacity to handle dynamic objective functions. Our research findings have been practically demonstrated across a spectrum of industrial applications, including tasks such as pick and place, welding, and gluing, and a unique triangle trajectory. Through real-world validation and optimization, our methodology has showcased its adaptability and versatility. The outcomes of our optimization efforts have yielded substantial benefits, resulting in 25% reduction in energy consumption and 12% decrease in cycle time when compared to average random values for robot TCP velocity and acceleration. Given the absence of industry-specific reference data, we conducted these comparisons with averaged random values, underscoring the effectiveness of our approach.

Share

COinS