Human-robot collaboration disassembly planning for end-of-life product disassembly process

Highlights

• A method for human-robot collaboration disassembly planning is proposed to improve the disassembly process efficiency of end-of-life products.

• New optimization parameters for human-robot collaboration based on the component target and operator change are defined and investigated.

• Experiments are carried out to validate the task classification and allocation for human-robot collaboration.

• An automotive case study is selected to test the proposed method, which results show a higher efficiency and reliability of the method in comparison with the particle swarm algorithm.

Abstract

The disassembly process is the main step of dealing with End-Of-Life (EOL) products. This process is carried out mostly manually so far. Manual disassembly is not efficient economically and the robotic systems are not reliable in dealing with complex disassembly operations as they have high-level uncertainty. In this research, a disassembly planning method based on human-robot collaboration is proposed. This method employs the flexibility and ability of humans to deal with complex tasks, alongside the repeatability and accuracy of the robot. Besides, to increase the efficiency of the process the components are targeted based on the remanufacturability parameters. First, human-robot collaboration tasks are classified, and using evaluation of components remanufacturability parameters, human-robot collaboration definition and characteristics are defined. To target the right components based on their remanufacturability factors, the PROMETHEE II method is employed to select the components based on Cleanability, Reparability, and Economy. Then, the disassembly process is represented using AND/OR representation and the mathematical model of the process is defined. New optimization parameters for human-robot collaboration are defined and the genetic algorithm was modified to find a near-optimal solution based on the defined parameters. To validate the task classification and allocation, a 6-DOF TECHMAN robot arm is used to test the peg-out-hole disassembly operation as a common disassembly task. The experiments confirm the task classification and allocation method. Finally, an automotive component was selected as a case study to validate the efficiency of the proposed method. The results in comparison with the Particle Swarm algorithm prove the efficiency and reliability of the method. This method produces a higher quality solution for the human-robot collaborative disassembly process.

Introduction

This research aims to solve the selective sequence planning problem for collaborative human-robot disassembly tasks. Disassembly sequence optimization is considered a challenging area of the robotic disassembly process. This difficulty is intensified when the process is designed for human-robot collaboration. Both human and robot can handle the tasks individually and interact with each other in many ways [1]. These different possibilities increase the possible optimum solutions exponentially [2]. To solve this problem, a new targeted disassembly strategy for human-robot collaboration is proposed.

Disassembly activities are considered as a new resource production for raw materials and manufacturing activity to produce components and products. The fourth industrial revolution which involves the digital revolution or as it is known as industry 4.0, is progressing rapidly. This activity is vital for implementing and enabling industry 4.0. The essentiality of the disassembly has been established; however, environmental issues and efficiency problems are considered as the bottleneck of this industry [3,4]. Emerging robots and especially cobots (collaborative robots) in the production industry have opened new doors in remanufacturing and disassembly process [5,6]. Unlike assembly, disassembly activities contain Haigh level uncertainty due to many factors. These factors include used and damaged products, operation failure, operation difficulty uncertainties, etc. The uncertainty and complexity of the process rise dramatically when robot and human are in collaboration. Liu et al. investigated the disassembly process using a human-robot collaborative system. They presented a systematic framework and studied the foundations of collaborative planning [7]. Task allocation and robotic disassembly of end-of-life (EOL) products were studied by Alshibli et al. [8]. They employed Taguchi's orthogonal arrays to investigate the disassembly optimization problem. Li et al. considered human fatigue in a human-robot collaborative disassembly process and investigated the sequence planning [9]. They studied disassembly task assignment in a collaborative environment and proposed a model for disassembly sequence planning using the Bees algorithm based on the human fatigue parameter. A human-robot disassembly system seems to be an optimum solution compare to manual systems or robotic disassembly [10,11]. The reason for this also can be due to high-level uncertainty and unpredicted operations, and different levels of difficulty of disassembly operation. Despite the potential benefit of the human-robot collaboration for disassembly activities, coordinated and optimum collaboration, and scheduling their tasks are challenging [12,13].

Based on Sanches and Haas [14], very few studies were carried out in the field of selective disassembly especially for the reuse of buildings structures. In fact, a brief literature search shows that there is a lack of research and analyses in this field and an established methodology has not been developed. To address this gap, Sanches et al. proposed a multi-objective method for optimization of selective disassembly planning using combining various de-construction approaches [15].

In order to generate a model that represents the disassembly precedence and interface relationships amongst the product components and subassemblies different methods were employed. A representation method should be able to present the precedence and interface relationship mathematically so the optimization method could be employed. Also, it should be able to model the disassembly constraints amongst the components and subassemblies [16]. Based on this definition there are four different categories for product disassembly representation: Graph-based representation, Petri nets, Matrix-based representation, and a mixture of these models [17]. Bourjault proposed disassembly trees as a basic graph-based disassembly representation [18]. The representation which is a combination of nodes is started with a root node representing the EOL product and continuing with branches that represents the components and sub-assemblies [19].

Robotics in the industry is growing sharply and its global market expanding significantly. Global sales of different robots have seen a 30% increase in 2017 which was rising for five consecutive years. Improvement in collaborative robotics combined with decreasing the price of industrial robots leads to further use and popularity of the robots in the future. In addition, as sensor prices falling rapidly, the Internet of Things (IoT) and industry 4.0 are emerging in the remanufacturing industry [20]. The fourth industrial revolution i.e. intelligent automation, which integrates robotics and its accessories to infrastructures and factories' cyber-physical system makes disassembly automation inevitable in the future. Also, Artificial Intelligence (AI) improvements in combination with adaptable end effectors provide more flexibility in disassembly automation. Implementation of automation and robotics was described by The World Economic Forum as a positive improvement for the world's employment and jobs in the future. It argued that this development could lead to the machine and human collaboration where they could complement each other. In order for the companies working in EOL product treatment to remain competitive in the remanufacturing market, successful automation in the future is vital. Remanufacturing in general and the disassembly process industry in particular, will not be able to survive in the future without an efficient automation strategy [21]. Several automated methods for EOL product disassembly have been investigated and proposed to increase the recovery of valuable components and materials. Li et al. utilised a robotic disassembly system to increase the recovery of important materials from electric vehicles [22]. They showed that using a robotic disassembly system a significant recovery value can be achieved. Different disassembly methods were investigated by Duou et al. [23] and the importance of autonomous disassembly systems was highlighted. Knoth et al. [24] employed a vision system to detect product components for disassembly and proposed an intelligent automated disassembly system. Vongbunyong et al. [25] studied the automated disassembly process of TV screens and employed a vision system to develop cognitive robotic systems. They showed that further configurability and more flexibility for robotic disassembly systems can be achieved.

The aim of disassembly planning is to use product representation information as the solution domain and plan the disassembly process i.e. finding the optimum disassembly sequence [26]. Based on the product representation method different DP methodologies have been developed [27]. As the nature of the disassembly planning problems is NP-hard type these methods are based on the heuristic algorithms [28]. In this method to find a near-optimal sequence, different variables such as cost, time, and disassembly efficiency have been studied as the objective of the planning i.e. optimization [29]. Parsa et al. defined a new disassemblability parameter based on the geometry of the components and the difficulty of the operation to optimise the disassembly process [30]. The proposed method by Parsa et al. tries to solve the disassembly sequence planning for manual disassembly based on the disassemblability characteristics of the components. However, this research is focused on human-robot disassembly sequence planning problem. In order to search and find a near-optimal solution for the assembly process, Pedrazzoli et al. introduced a set of rules [31]. Zhang et al. used search rules to study the operation plan of robot gripper [32]. Objectives of the disassembly planning e.g. higher profit, lower environmental impact, lower cost, feasibility, etc. tend to define the aim of disassembly planning. Based on the different purposes of the disassembly process, specific objectives are determined. The main categories of disassembly objectives are environmental objectives, process cost, revenue, and other objectives. Each category has many sub-categories that were studied by different studies. For instance, many researchers investigated process time, cost of disassembly tools, operation, and part numbers as part of cost objective [33].

Tasks scheduling and finding an optimum sequence of operations for both human and robot to work together have a direct effect on the process efficiency. Therefore, in this work, finding an optimum solution by which human and robot could be able to collaborate efficiently is explored. In addition, the remanufacturability of a component is considered an important factor to make the disassembly process profitable economically.

The contribution of the present work is threefold. First, a new quantitative scoring algorithm is proposed to evaluate each disassembly operation task quantitatively. This, score-based evaluation, is employed to classify and allocate a disassembly task to human or robot. Secondly, for the first time, the remanufacturability factor is considered for human-robot collaboration task optimisation. The Remanufacturability of a component is an important parameter that determines the profit of the entire process. Finally, the classification and allocation of a disassembly task using score-based evaluation are evaluated based on empirical experiments using a collaborative robot, and results are used to validate the proposed method.

Therefore, in this research first, new remanufacturability parameters are defined to select targeted components. Then, the principles of human-robot collaboration and its characteristics are defined and analysed. The operation tasks are classified based on newly defined parameters for human-robot collaboration and the disassembly problem is mathematically presented. Genetic algorithm is employed and modified to search the solution domain for a near-optimal solution. Finally, the proposed method is examined on the real automotive component to validate and test its efficiency.