Abstract: Aiming at the problem that the working posture of virtual human was complicated to simulate and lacked verisimilitude,a working posture simulation method for virtual human based on multi-objective genetic algorithm was presented.Based on the mapping relationships between objective functions of ergonomic factors and working posture parameters,the multi-objective optimization problem of working posture simulation for virtual human was established.The multi-objective genetic algorithm was adopted to solve this optimization problem and obtain the Pareto optimal solutions.The optimal solutions were sorted by the weight of each ergonomic factor to find the most appropriate posture parameters.Case studies showed that the proposed method could generate realistic working posture of virtual human and has good adaptability in a variety of virtual environments.

Key words: virtual human, working posture, simulation, multi-objective optimization, genetic algorithm, ergonomic factors

摘要： 针对虚拟人作业姿态仿真操作复杂和仿真所得姿态逼真性不足的问题,提出一种基于多目标遗传算法的虚拟人作业姿态仿真方法。建立了作业环境中虚拟人的人机因素目标函数与姿态参数的映射关系,在此基础上构建了面向虚拟人作业姿态仿真的多目标优化问题。通过多目标遗传算法计算得出该虚拟人作业姿态参数优化问题的Pareto最优解集,根据各人机因素权重对解集排序并挑选最优值对应的姿态参数方案作为最佳姿态解。仿真结果表明,该方法可以生成自然逼真的虚拟人作业姿态,在多样化的作业环境下表现出良好的适用性。

关键词: 虚拟人, 作业姿态, 仿真, 多目标优化, 遗传算法, 人机因素