Search results

As complex analysis of contact models is required in the traditional assembly strategy, it is still a challenge for a robot to complete the multiple peg-in-hole assembly tasks autonomously. This paper aims to enable the robot to complete the assembly tasks autonomously and more efficiently, with the strategies learned by reinforcement learning (RL), a learning-accelerated deep deterministic policy gradient (LADDPG) algorithm is proposed.

The multiple peg-in-hole assembly strategy is designed in two modules: an advanced planning module and a bottom control module. The advanced module is completed by the LADDPG agent, which is used to derive advanced commands based on geometric and environmental constraints, that is, the desired contact force. The bottom-level control module will drive the robot to complete the compliant assembly task through the adaptive impedance algorithm according to the command set issued by the advanced module. In addition, a set of safety assurance mechanisms is developed to safely train a collaborative robot to complete autonomous learning.

The method can complete the assembly tasks well through RL, and it can realize satisfactory compliance of the robot to the environment. Compared with the original DDPG algorithm, the average values of the instantaneous maximum contact force and contact torque during the assembly process are reduced by approximately 38% and 74%, respectively.

The entire algorithm can also be applied to other robots and the assembly strategy can be applied in the field of the automatic assembly.

A compliant assembly strategy based on the LADDPG algorithm is proposed to complete the automated multiple peg-in-hole assembly tasks.

This paper aims to present an optimization algorithm combined with the impedance control strategy to optimize the robotic dual peg-in-hole assembly task, and to reduce the assembly time and smooth the contact forces during assembly process with a small number of experiments.

Support vector regression is used to predict the fitness of genes in evolutionary algorithm, which can reduce the number of real-world experiments. The control parameters of the impedance control strategy are defined as genes, and the assembly time is defined as the fitness of genes to evaluate the performance of the selected parameters.

The learning-based evolutionary algorithm is proposed to optimize the dual peg-in-hole assembly process only requiring little prior knowledge instead of modeling for the complex contact states. A virtual simulation and real-world experiments are implemented to demonstrate the effectiveness of the proposed algorithm.

The proposed algorithm is quite useful for the real-world industrial applications, especially the scenarios only allowing a small number of trials.

The paper provides a new solution for applying optimization techniques in real-world tasks. The learning component can solve the data efficiency of the model-free optimization algorithms.

In the future, large space truss structures will be likely to require on-orbit assembly. One of the several proposed methods includes cooperative assembly performed by pressure-suited astronauts during extravehicular activity (EVA) and space robots. An intelligent planning method was presented to generate optimal assembly tasks.

Firstly, the inherent hierarchical nature of truss structures allows assembly sequences to be considered from strut level and structural volume element (SVE) level. Then, a serial assembly strategy in human-robot environment was applied. Furthermore, a two-level planning algorithm was presented. At the first-level planning, one ant colony algorithm for assembly sequence planning was improved to adopt assembly direction and time as heuristic information and did not consider assembly tasks. And, at the second-level planning, another novel colony algorithm for assembly task planning mainly considered results of the first-level planning, human-robot interactive information, serial assembly strategy and assembly task distributions.

The proposed two-level planning algorithm is very effective for solving the human and robot cooperative assembly of large space truss structures.

In this paper, the case study is based on the following assumptions: each tetrahedron is assembled by two astronauts; each pentahedron is assembled by three astronauts.

A case illustrates the results of the two-level planning. From this case study, because of geometrical symmetry nature of large space truss structures, the optimal assembly sequences are not only one.

The improved ant colony algorithm can deal with the assembly sequence and task planning in human-robot environment more effectively.

The paper aims to propose an assembly scheme based on master–slave coordination for a compliant dual-arm robot to complete a peg-in-hole assembly task.

The proposed assembly scheme is inspired by the coordinated behaviors of human beings in the assembly process. The left arm and right arm of the robot are controlled to move alternately. The fixed arm and the moving arm are distinguished as the slave arm and the master arm, respectively. The position control model is used at the uncontacted stage, and the torque control model is used at the contacted stage.

The proposed assembly scheme is evaluated through peg-in-hole assembly experiments with different shapes of assembly piece. The round, triangle and square assembly piece with 0.5 mm maximum clearance between the peg and the hole can be assembled successfully based on the proposed method. Furthermore, three assembly strategies are investigated and compared in the peg-in-hole assembly experiments with different shapes of assembly piece.

The contribution of this study is that the authors propose an assembly scheme for a compliant dual-arm robot to overcome the low positioning accuracy and complete the peg-in-hole assembly tasks with different shapes parts.

To provide a new model to solve the assembly‐planning problem of a textile machine in a shopfloor which can help researchers and practitioners.

The assembly planning of a textile machine (repetitive manufacturing system) involves the allocation of operations to cross‐trained operators. Workflow is defined as the workloads assigned to the operators. Operators with smaller workloads are selected to be assigned new operations from the list of unscheduled operations. Three different scheduling strategies – random, shortest processing time, and longest processing time – are adopted for the selection of operations to be assigned to operators. Different combinations of these strategies are considered for the selection of both preceding and succeeding operations. A computer simulation program has been coded on an IBM/PC‐compatible system in the C++ language to study the performance of real data from the shopfloor.

The relative percentage of imbalance is adopted for evaluating the performance of these heuristics. The RL, SL and LL produced well balanced workload schedules with lesser RPI values for all operators other than heuristics.

Non‐traditional approaches like genetic algorithms can be applied to determine the robustness of the results obtained by this research.

The experiments on simulated and real data clearly indicate that the order of succeeding operations determines the balanced workflows to the assembly of operations among the operators.

The allocation of assembly operations to the operators is modeled into a parallel machine‐scheduling problem with precedence constraints using the objective of minimizing the workflow among the operators.

This study aims to present a new haptic-enabled virtual assembly system for the automatic generation and objective assessment of assembly plans. The system is intended to be used as an assembly planning tool along the product development process.

The generation of product assembly plans is based on the analysis of the assembly movements and operations performed by the user during the virtual assembly execution, and the objective assessment of product assembly is based on the definition and computation of new proposed assembly metrics.

To evaluate the system, a case study corresponding to the assembly of a mechanical component is presented and analyzed. The results demonstrate that the proposed system is an effective tool to plan and evaluate different product assembly strategies in a more practical and objective approach than existing assembly planning methods.

Although the virtual assembly execution time is larger than the real assembly execution time, the assembly planning and evaluation results provided by the system are valid. However, the development of higher performance collision detection algorithms is needed to reduce the simulation time.

The proposed virtual assembly system is able to not only simulate and automatically generate assembly plans but also objectively assess them from the virtual assembly task execution. The introduction and use of several assembly performance metrics to objectively evaluate assembly strategies in virtual assembly also represents a novel contribution.

The paper aims to propose a method to build environmental constraint region online in complex-shaped peg-in-hole assembly tasks.

Compared with conventional way which using computer-aided design (CAD) models of assembly parts to construct the environmental constraint region offline, the paper provides an online approach that consists of three aspects: modeling assembly parts through visual recognition, decomposing complex shapes into multiple primitive convex shapes and a numerical algorithm to simulate the peg-in-hole insertion contact. Besides, a contrast experiment is performed to validate the feasibility and effectiveness of the method.

The experiment result indicates that online construction takes less time than the offline way under the same task conditions. Furthermore, due to the CAD models of the parts are not required to be known, the method proposed in the paper has a broader application in most assembly scenarios.

With the improvement of customization and complexity of manufactured parts, the assembly of complex-shaped parts has drawn greater attention of many researchers. The assembly methods based on attractive region in environment (ARIE) have shown great performance to achieve high-precision manipulation with low-precision systems. The construction of environmental constraint region serves as an essential part of ARIE-based theory, directly affect the formulation and application of assembly strategies.

Examines the conceptual design of robotic assembly systems in conjunction with the analysis and optimization of the product and process design. Explains how an integral assembly model is utilised to study the relationships between assembly variables which play a role in each stage of the design process. Outlines the characteristics and benefits of the concentric design process and explains the total productivity concept. Concludes that the integral assembly model, which underlies the concentric design process, provides the opportunity to store product, process and system data and can therefore be considered as a reference model for product development and process planning as well as for the design and analysis of assembly systems.

Market turbulence forces assembly plants to constantly adjust their production volume of products, variants and quantities. At the same time, assembly plant managers must protect long‐term investments in the flexible assembly system. For reconfigurability and agility the best solution is the modular semi‐automatic approach by combining flexible automation and human skills. It gives managers possibility to adjust volume by adding new modules or to automate the manual tasks step by step. The control of material handling and information flow in the agile assembly system is important. To keep flexibility, the combination of an intelligent pallet, i.e. use of escort memory, carrying a single product together with other hardware providing paperless production even supports a lot size of one. The article shows how to create flexible capability and capacity in the final assembly systems.

This paper develops the statistical error analysis model for assembling, to derive measures of controlling the geometric variations in assembly with multiple assembly stations, and to provide a statistical tolerance prediction/distribution toolkit integrated with CAD system for responding quickly to market opportunities with reduced manufacturing costs and improved quality. First the homogeneous transformation is used to describe the location and orientation of assembly features, parts and other related surfaces. The desired location and orientation, and the related fixturing configuration (including locator position and orientation) are automatically extracted from CAD models. The location and orientation errors are represented with differential transformations. The statistical error prediction model is formulated and the related algorithms integrated with the CAD system so that the complex geometric information can be directly accessed. In the prediction model, the manufacturing process (joining) error, induced by heat deformation in welding, is taken into account.