Study On Physics-based Virtual Assembly Technology

Product assembly is the most important and laborious step during the product development process and is the last step of achieving whole product performance. It has serious influences on production quality, development cycle and production cost. The general CAD software neglects the assembly process, which can only move the assembling part to the final position according to the constraints between parts, thus it can not simulate the collisions between parts and verify the feasibility of assembly path. Using virtual reality technologies to establish a real assembly environment on a virtual prototype. The operator can conduct an assembly process simulation corresponding to the real world. Based on the realistic assembly simulation, the assemblability of parts, the reasonability of assembly spatial arrangement, the reachability of assembly operation and the comfort of operator are verified. Using the virtual assembly environment to evaluate the product assembly performance and assemblability in the early product design stage, find out the potential collisions during assembly process and the design deficiencies, and then prompt the designers to find and solve the problems as early as possible. Thus, it will optimize product assembly performance, improve the quality of products, shorten the development cycle and enhance the market competitiveness.This paper aims at constructing a series of assembly simulation and assemblability evaluation approaches based on physical attributes and realizing high consistent of virtual and real assembly process and the quantitative evaluation of assemblability. The relevance theories and key technologies mainly include physics-based virtual assembly system architecture, virtual product modeling, physics-based assembly process simulation, virtual assembly guidance, collision detection, human factors analysis and product assemblability evaluation.The main content of this dissertation is summarized as follows.1. The functional requirements of physics-based virtual assembly system are analyzed based on the description of research objectives and research thinking of physics-based virtual assembly. The architecture of physics-based virtual assembly is established, the workflow of the system is given and the hierarchical structure and research contents in each stage of the system are analyzed in detail. The key technologies involved in the system are divided into the foundational techniques and the functional techniques. At last, the key technologies are described briefly.2. The physics-based virtual models are created by analyzing the requirements of physical attributes involved in the virtual parts. The data structure and the program expression of the virtual models are proposed. The calculation methods of basic physical attributes (eg. rotational inertia, center of mass) are given. The dynamics equations and kinematic equations of multi-rigid-body systems under virtual assembly environment are built. A collision detection method based on adaptive test lines is presented, solving the problem of penetration and crossing in collision detection for high speed objects. The methods of test lines construction and crossing-frame processing are proposed. A case study is used to analyze the detection accuracy and efficiency of the method. The results show that this method guarantees a high detection accuracy as well as improves the detection efficiency by translating the the intersection calculation between bodies into that between body and lines.3. The virtual assembly process is divided into early assembly stage and later assembly stage. The problem of motion navigation and position in later assembly stage is studied. A motion navigation method based on force guidance is presented. The assembling part is positioned under the combined action of external forces and moments. At each step, the dynamic equations are calculated to obtain the motion parameters and position response of the assembling part. The Newton-Euler equations used for dynamic analysis are established and the calculation methods of motion parameters and pose transformation matrix are given. The calculation methods of assembly force and assembly torque considering human factors and uncertainties during assembly process are presented. Using the Monte Carlo method to estimate the target position of the assembling part reflects the intelligence and fuzziness of motion navigation. The calculation method of contact force based on collision response is proposed, which avoids the collided objects crossing each other and ensures the reality of assembly.4. The basic principles and implementation methods of particle filter technology are studied. The basic idea and implementation process of Monte Carlo method, importance sampling method and resampling method are discussed. The principles and steps of MCMC particle filter algorithm using MH sampling are given. Based on these basic algorithms, an assembly guidance method based on particle filter is presented, solving the problem that the existing methods can not conduct the assembly task automatically and show the realistic assembly process. The workflow and detailed pose transform steps are given. The influences of the number of samples and the parts shapes on the performance of this method and assembly efficiency are discussed.5. A human factors-oriented product assemblability evaluation method is presented. The quantitative calculation method of human factors during assembly process is given. The influences of human factors on assembly performance are analyzed. The component assemblability evaluation system based on physical assembly simulation is established. The influence of structure factor on component assemblability is analyzed. From an overall perspective, a product assemblability evaluation method is presented on the basis of assemblability of each component and assembly sequence. This method is based on the realistic assembly process simulation. Thus, compared with the traditional assemblability evaluation methods, the evaluation results of the proposed method have theoretical basis and are more accurate and reliable.6. A physics-based virtual assembly system-PVAS is developed based on above methods and key technologies. The running and developing environment of system is introduced and the functional model of system is constructed. Some test cases are carried out to test and verify the performance and capabilities of the proposed system and the relevant methods.