多维曲面成形运动规划中三维模型数据处理研究
Research on 3D Model Data Processing for Multi-Dimensional Surface Forming Motion Planning
胡天雄 1王绍宗 1郭智 1冉跃龙 1赵海波1
作者信息
- 1. 中国机械科学研究总院先进成形技术与装备全国重点实验室,北京 100083
- 折叠
摘要
多维曲面成形过程中三维模型读取和空间目标点的求解效率对提高成形速度具有重要影响作用.随着STL模型复杂度提升,三角面片数量相应增加,计算复杂度大幅增长,求解效率降低,运动规划和图案映射过程计算难度也随之增加.针对上述问题基于深度学习方法提出一种三维模型数据处理及运动规划方法,并与传统的计算规划方法进行对比.实验结果表明,所提方法能有效提升曲面打印中模型数据点定位速度2 倍以上,定位速度随求解点数据量增加而提升,且对模型顶点数增加具有鲁棒性,大大提升运动规划速度,为多维曲面成形运动规划求解提供新思路,适用于大规模数据计算且可被用于直接交互现实的生产制造任务.
Abstract
The efficiency of reading 3D models and solving spatial target points during the process of multi-dimensional surface forming is crucial.As the complexity of the STL model increases,the number of trian-gular patches increases correspondingly,the computational complexity increases significantly,the solution's efficiency decreases,and the computational difficulty of motion planning and pattern mapping also increa-ses.This paper proposes a 3D model data processing and motion planning method based on deep learning method for the above problems and compares it with the traditional computational planning method.The ex-perimental results show that the method proposed in this paper can effectively improve the positioning speed of model data points by more than 2 times,and the positioning speed increases with the increase of the a-mount of solved point dataand it is robust to the increase of the number of model vertices,which greatly im-prove the speed of motion planning and provide a new idea for solving the motion planning of multi-dimen-sional surface forming.It is suitable for large-scale data calculation and can be applied to production and manufacturing tasks interacting with reality directly in the future.
关键词
多维曲面成形/深度学习/三维模型数据处理/运动规划Key words
multi-dimensional surface forming/deep learning/3D model data processing/motion planning引用本文复制引用
基金项目
北京市概念验证平台项目(20220481100)
出版年
2024
NETL
NSTL
维普
万方数据