基于DMPs的双机器人协同搬运轨迹的学习与泛化

doi:10.12141/j.issn.1000-565X.230013

华南理工大学学报(自然科学版) ›› 2023, Vol. 51 ›› Issue (12): 9-20.doi: 10.12141/j.issn.1000-565X.230013

所属专题： 2023年机械工程

基于DMPs的双机器人协同搬运轨迹的学习与泛化

曹学鹏¹ 王德硕¹^,² 冯艳丽² 樊豪¹^,² 张弓³ 张新荣¹ 赵睿英¹

^1.长安大学机电液一体化研究所，陕西西安 710064
^2.西安航天时代精密机电有限公司，陕西西安 710100
^3.广州中国科学院先进技术研究所，广东广州 511458

收稿日期:2023-01-10 出版日期:2023-12-25 发布日期:2023-04-21
通信作者: 冯艳丽（1984-），女，博士，研究员，主要从事关节和移动机器人控制系统技术研究。 E-mail:115916192@qq.com
作者简介:曹学鹏（1982-），男，博士，教授，主要从事机电控制系统及工业机器人应用研究。E-mail： tiepeng2001@chd.edu.cn
基金资助:
国家自然科学基金面上项目(62073092);陕西省重点研发计划项目(2021ZDLGY09-02)

Learning and Generalization of Dual-Robot Cooperative Handling Trajectory Based on Dynamic Movement Primitives

CAO Xuepeng¹ WANG Deshuo¹^,² FENG Yanli² FAN Hao¹^,² ZHANG Gong³ ZHANG Xinrong¹ ZHAO Ruiying¹

^1.Institute of Mechatronics and Hydraulics，Chang’an University，Xi’an 710064，Shaanxi，China
^2.Xi’an Aerospace Times Precision Electromechanical Company Limited，Xi’an 710100，Shaanxi，China
^3.Guangzhou Institute of Advanced Technology，Chinese Academy of Science，Guangzhou 511458，Guangdong，China

Received:2023-01-10 Online:2023-12-25 Published:2023-04-21
Contact: 冯艳丽（1984-），女，博士，研究员，主要从事关节和移动机器人控制系统技术研究。 E-mail:115916192@qq.com
About author:曹学鹏（1982-），男，博士，教授，主要从事机电控制系统及工业机器人应用研究。E-mail： tiepeng2001@chd.edu.cn
Supported by:
the General Program of National Natural Science Foundation of China(62073092);the Key R&D Plan Projects in Shaanxi Province(2021ZDLGY09-02)

摘要/Abstract

摘要：

针对双机器人协作系统执行具有强协调约束关系的仿人任务时，存在轨迹学习复杂、协调约束分析欠缺的问题，提出了基于动态运动基元（DMPs）的双机器人协同搬运轨迹学习及泛化方法。首先，从双机器人协同搬运任务出发，分析了双机器人协调约束关系，建立了双机器人运动约束模型。然后，将机器人运动轨迹解耦为位置轨迹和姿态轨迹，采用四元数实现姿态轨迹的无奇异描述，分别建立位置轨迹和姿态轨迹的动态运动基元模型，结合双机器人运动约束模型与动态运动基元模型，兼顾各自的任务要求和相对位姿约束，进而获得双机器人的运动轨迹。接着，开展了双机器人协同搬运轨迹的仿真与实验，结果表明：采用双机器人协同搬运轨迹的学习与泛化方法，当改变起、终点状态时，双机器人定姿态协同搬运的起、终点位置误差分别为0.029 2 mm和0.112 7 mm，变姿态协同搬运的起、终点位置误差分别为0.032 3 mm和0.113 1 mm，终点的四元数姿态偏差为0.001 4、0.002 7、0.001 8、0.003 0，表明该协同搬运轨迹的学习与泛化方法具有较高的运动控制精度，即使起、终点任务参数改变，泛化轨迹仍可保证目标的可达性，验证了提出的双机器人协调运动轨迹控制方法的合理性和有效性。该方法可学习人体搬运过程并能够准确泛化出新的运动轨迹，实现了双机器人的协调运动，具有重要的工程应用价值。

关键词: 动态运动基元, 双机器人协同, 轨迹学习, 协调约束, 四元数

Abstract:

Aiming at the problems of complex trajectory learning and lack of coordination constraint analysis when a dual-robot collaborative system performs humanoid tasks with strong coordination constraints, this paper proposed a dual-robot cooperative handling trajectory learning and generalization method based on dynamic movement primitives (DMPs). Firstly, starting from the dual-robot cooperative handling task, the coordination constraints of the dual-robot were analyzed, and the motion constraint model of the dual-robot was established. Then, the robot motion trajectory was decoupled into position trajectory and orientation trajectory, and the quaternion was used to realize the non-singular description of the orientation trajectory. And the dynamic movement primitives model of position trajectory and orientation trajectory were established respectively. They were combined with the dual robot motion constraint model and DMPs model, and the dual-robot movement trajectory was obtained, taking into account their respective task requirements and relative pose constraints. Finally, the simulation and experiments of the cooperative handling trajectory of the two robots were carried out. The results show that: using the learning and generalization method of the dual-robot cooperative handling trajectory, when the starting and ending states are changed, the position errors of start point and end point of the dual-robot cooperative handling with the fixed orientation are 0.029 2 mm and 0.112 7 mm respectively; the position errors of start point and end point of variable orientation coordinated handling are 0.032 3 mm and 0.113 1 mm respectively; and the quaternion orientation errors of the end point are 0.001 4, 0.002 7, 0.001 8, 0.003 0, indicating that the cooperative handling trajectory learning and generalization method has high motion control accuracy; even if the task parameters of the starting and ending are changed, the generalized trajectory can still ensure the accessibility of the target, which verified the scientificity and effectiveness of the proposed dual-robot coordination motion trajectory control strategy. The method proposed in this paper can effectively learn the human handling process and can accurately generalize new motion trajectories. It realizes the dual-robot coordinated motion and has important engineering application value.

Key words: dynamic movement primitive, dual-robot collaboration, trajectory learning, coordination constraint, quaternion

中图分类号:

TP242.2

曹学鹏, 王德硕, 冯艳丽, 等. 基于DMPs的双机器人协同搬运轨迹的学习与泛化[J]. 华南理工大学学报(自然科学版), 2023, 51(12): 9-20.

CAO Xuepeng, WANG Deshuo, FENG Yanli, et al. Learning and Generalization of Dual-Robot Cooperative Handling Trajectory Based on Dynamic Movement Primitives[J]. Journal of South China University of Technology(Natural Science Edition), 2023, 51(12): 9-20.

图/表 18

图1

图2

图3

图4

图5

表1

三维轨迹学习与泛化最大终点位置误差 (mm)"

轨迹形式	$Δ x, Δ y, Δ z$	空间误差 $ε$
仅改变起点	0.008 5，0.007 9，0.008 8	0.014 6
仅改变终点	0.005 6，0.005 6，0.007 3	0.010 8
起、终点均改变	0.012 1，0.009 4，0.012 6	0.019 8

表1

图6

图7

图8

表2

图9

图10

图11

图12

表3

定姿态搬运实验的起、终点位置误差"

参数	$Δ x$ /mm	$Δ y$ /mm	$Δ z$ /mm
起点误差	0.017	0.021	0.011
终点误差	0.044	0.095	0.039

表3

图13

图14

表4

变姿态搬运实验的起、终点状态偏差"

参数	$Δ x$ /mm	$Δ y$ /mm	$Δ z$ /mm	$Δ q w$	$Δ q x$	$Δ q y$	$Δ q z$
起点误差	0.022	0.019	0.014	0.000 3	0.000 6	0.001 1	0.000 4
终点误差	0.025	0.102	0.042	0.001 4	0.002 7	0.001 8	0.003 0

表4

参考文献 26

1	JI W， WANG L ．Industrial robotic machining：a review［J］．The International Journal of Advanced Manufacturing Technology，2019，103（1）：1239-1255.
2	BIEN Z， LEE J ．A minimum-time trajectory planning method for two robots［J］．IEEE Transactions on Robotics and Automation，1992，8（3）：414-418.
3	PETAR C， BOJAN J ．Dual-arm robot motion planning based on cooperative coevolution［J］．Emerging Trends in Technological Innovation，2010，314（6）：169-178.
4	SMITH C， KARAYIANNIDIS Y， NALPANTIDIS L，et al ．Dual arm manipulation-a survey［J］．Robotics and Autonomous Systems，2012，60（10）：1340-1353.
5	GAN Y H， DAI X ．Kinematic cooperation analysis and trajectory teaching in multiple robots system for welding［C］∥Proceedings of ETFA 2011．Toulouse：IEEE，2011：1-8.
6	LI J， LIU Y， ZANG X ．Constraints analysis and motion planning for coordinated manipulation of a dual-arm robot［C］∥Proceedings of 2018 IEEE International Conference on Information and Automation （ICIA）．Wuyi Mountain：IEEE，2018：1422-1426.
7	GULLETTA G， ERLHAGEN W， BICHO E ．Human-like arm motion generation：a review［J］．Robotics，2020，9（4）：102-149.
8	SCHAAL S ．Is imitation learning the route to humanoid robots？［J］．Trends in Cognitive Sciences，1999，3（6）：233-242.
9	KOENIG N， MATARIC M ．Robot life-long task learning from human demonstrations：a Bayesian approach［J］，Autonomous Robots，2017，41（5）：1173-1188.
10	PIGNAT E， CALINON S ．Learning adaptive dressing assistance from human demonstration［J］．Robotics and Autonomous Systems，2017，93：61-75.
11	SHIN S Y， KIM C H ．Human-like motion generation and control for humanoid’s dual arm object manipulation［J］．IEEE Transactions on Industrial Electronics，2014，62（4）：2265-2276.
12	QU J， ZHANG F， WANG Y，et al ．Human-like coordination motion learning for a redundant dual-arm robot［J］．Robotics and Computer-Integrated Manufacturing，2019，57：379-390.
13	IJSPEERT A J， NAKANISHI J， HOFFMANN H，et al ．Dynamical movement primitives：learning attractor models for motor behaviors［J］．Neural Computation，2013，25（2）：328-373.
14	COHEN Y， BARSHIR O， BERMAN S ．Motion adaptation based on learning the manifold of task and dynamic movement primitive parameters［J］．Robotica，2021，39（7）：1299-1315.
15	KULVICIUS T， BIEHL M， AEIN M J，et al ．Interaction learning for dynamic movement primitives used in cooperative robotic tasks［J］．Robotics and Autonomous Systems，2013，61（12）：1450-1459.
16	UMLAUFT J， SIEBER D， HIRCHE S ．Dynamic movement primitives for cooperative manipulation and synchronized motions［C］∥Proceedings of 2014 IEEE International Conference on Robotics and Automation （ICRA）．Hong Kong：IEEE，2014：766-771.
17	ZHAO T， DENG M， LI Z，et al ．Cooperative manipulation for a mobile dual-arm robot using sequences of dynamic movement primitives［J］．IEEE Transactions on Cognitive and Developmental Systems，2018，12（1）：18-29.
18	孟石，戴先中，甘亚辉．多机器人协作系统轨迹约束关系分析及示教方法［J］．机器人，2012，34（5）：546-552，565.
	MENG Shi， DAI Xian-zhong， GAN Ya-hui ．Path constraint relation and trajectory teaching method for multi-robot cooperation System［J］．Robot，2012，34（5）：546-552，565.
19	张磊，王威．双机器人协同搬运运动学分析及路径规划［J］．机械工程与自动化，2020（2）：89-91，94.
	ZHANG Lei， WANG Wei ．Kinematics analysis and path planning of dual-robot collaborative handling ［J］．Mechanical Engineering and Automation，2020（2）：89-91，94.
20	IJSPEERT A J， NAKANISHI J， SCHAAL S ．Movement imitation with nonlinear dynamical systems in humanoid robots［C］∥Proceedings of 2002 IEEE International Conference on Robotics and Automation．Washington，DC：IEEE，2002：1398-1403.
21	HOFFMANN H， PASTOR P， PARK D H，et al ．Biologically-inspired dynamical systems for movement generation：automatic real-time goal adaptation and obstacle avoidance［C］∥Proceedings of 2009 IEEE International Conference on Robotics and Automation．Kobe：IEEE，2009：2587-2592.
22	UDE A， NEMEC B， PETRIC T，et al ．Orientation in cartesian space dynamic movement primitives［C］∥Proceedings of 2014 IEEE International Conference on Robotics and Automation （ICRA）．Hong Kong：IEEE，2014：2997-3004.
23	张磊，方灶军，王聚幸，等．基于任务参数加权的动态运动基元泛化方法［J］．中国机械工程，2022，33（10）：1226-1233，1243.
	ZHANG Lei， FANG Zao-jun， WANG Ju-xing，et al ．Generalization method of dynamic movement primitives based on the weighting of task parameters［J］．China Mechanical Engineering，2022，33（10）：1226-1233，1243.
24	SOLA J ．Quaternion kinematics for the error-state Kalman filter［J］．arXiv preprint arXiv：，2017.
25	王健发，王耀南，陈文锐，等．基于改进动态运动基元的6D轨迹规划［J］．控制理论与应用，2022，39（5）：809-818.
	WANG Jian-fa， WANG Yao-nan， CHEN Wen-rui，et al ．6D trajectory planning based on improved dynamic motion primitives［J］．Control Theory and Applications，2022，39（5）：809-818.
26	工业机器人性能规范及其试验方法：［S］．

参数	x/mm	y/mm	z/mm	q_w	q_x	q_y	q_z
初始状态	-250	-310	200	0.959 7	0.176 8	0.128 4	-0.176 8
目标状态	-310	270	270	0.498 2	0.613 8	0.135 2	-0.597 9

[1]	翟敬梅, 路东伟. 按摩机器人优化示教策略及BPNN-DMPs轨迹学习模型[J]. 华南理工大学学报(自然科学版), 2023, 51(12): 1-8.
[2]	王丹丹谭开拓高素玲袁赣南. 基于模糊逻辑四元数的平方根 UKF 算法[J]. 华南理工大学学报(自然科学版), 2019, 47(4): 53-60.
[3]	陶玲王跃钱志余. 基于单位四元数的手术导航注册方法[J]. 华南理工大学学报（自然科学版）, 2011, 39(9): 128-133.
[4]	肖健文福拴. 基于协调约束的混合电力市场阻塞调度[J]. 华南理工大学学报（自然科学版）, 2009, 37(7): 112-118.

基于DMPs的双机器人协同搬运轨迹的学习与泛化

Learning and Generalization of Dual-Robot Cooperative Handling Trajectory Based on Dynamic Movement Primitives

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 18

参考文献 26

相关文章 4

编辑推荐

Metrics

本文评价