基于无迹卡尔曼改进微分动态轨迹规划的实时算法  

Improved real-time differential dynamic trajectory planning algorithm based on unscented kalman filter

作　　者：刘伟 蔡晨星 江如海 马利强 李昊 Liu Wei;Cai Chenxing;Jiang Ruhai;Ma Liqiang;Li Hao(Institute of Machinery & Electrical Engineering, Anhui Jianzhu University, Hefei 230601, China;Hefei CAS-Juche Technology Co.Ltd., Hefei 230601, China)

机构地区：[1]安徽建筑大学机械与电气工程学院,合肥230601 [2]合肥中科智驰科技有限公司,合肥230601

出　　处：《黑龙江科技大学学报》2022年第3期406-411,共6页Journal of Heilongjiang University of Science And Technology

基　　金：安徽省重点研究与开发计划项目(202004a05020066)。

摘　　要：针对经典微分动态规划算法优化机器人移动轨迹实时性差的问题,根据拉格朗日插值法建立二阶动力学控制模型,采用无迹变换传递动力学参数与差分简化了海森矩阵和梯度计算方法,对比分析了无迹微分动态规划算法与原算法优化机器人轨迹的迭代时间。结果表明,改进算法的求解时间较原算法缩短了37.2%。无迹微分动态规划算法在轨迹最优规划中保留了原算法求解特性且提高了轨迹规划实时性。This paper is focused on a study in response to the poor real-time performance of the classical differential dynamic programming algorithm in optimizing the robot trajectory.The study involves developing a second-order dynamic control model based on the Lagrange interpolation method;simplifying the Hessian matrix and gradient calculation method by using the method of traceless transformation to transfer the dynamic information and difference;comparing and analyzing the iteration time of optimizing the robot trajectory in using the traceless differential dynamic programming algorithm and the original algorithm.The results show that the solution time with the improved algorithm is shortened by 37.2%in constrast with original algorithm.The traceless differential dynamic programming algorithm in the optimal trajectory planning retains the solving characteristics of the original algorithm and improves the real-time performance of trajectory planning tested with simulation and experiment.

关 键 词：轨迹规划 无迹变换 无迹微分动态规划 实时性 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]