林立民, 刘成菊, 马璐, 王德明, 陈启军. 基于STFT的仿人机器人手臂轨迹生成方法[J]. 机器人, 2019, 41(5): 591-600.DOI: 10.13973/j.cnki.robot.180784.
LIN Limin, LIU Chengju, MA Lu, WANG Deming, CHEN Qijun. Arm Trajectory Generation for Humanoid Robot Based on STFT. ROBOT, 2019, 41(5): 591-600. DOI: 10.13973/j.cnki.robot.180784.
摘要为了提高机器人轨迹生成算法的泛化性,提出了一种基于时间-空间特征模板(STFT,spatio-temporal feature template)的机器人手臂轨迹生成方法.首先,针对机器人示教轨迹往往存在的时间长短和幅度差异较大的问题,采用广义的典型时间规整(generalized canonical time warping,GCTW)方法来统一时间和幅度的变化,从而获取机器人示教轨迹的共同特征模板.其次,基于STFT,引入机器人轨迹生成的平滑性约束、任务约束等因素,设定轨迹生成的目标函数并优化.最终在NAO仿人机器人平台上验证了所提出的轨迹生成算法,基于STFT生成机器人弧形轨迹并完成数字书写.实验结果表明,本文提出的基于STFT的轨迹生成策略可以生成满足期望条件的机器人轨迹,并具有一定的泛化性.
Abstract:To improve the generalization performance of robot trajectory generation algorithms, a manipulator trajectory generation strategy based on spatio-temporal feature template (STFT) is proposed. Firstly, to solve the problem that the manipulation time and amplitude of the demonstration trajectories are different greatly, the generalized canonical time warping (GCTW) method is adopted to unify the variation of the time and amplitude. In this way, the common feature template for demonstration trajectories is obtained. Secondly, based on STFT, smoothness constraints and task constraints of robot trajectory generation are introduced into the objective function for trajectory generation, and the function is then optimized. Finally, the trajectory generation strategy is verified on NAO robot platform. The curved trajectories and digital figure trajectories are generated based on STFT. The experimental results show that the proposed STFT-based trajectory generation strategy can generate robot trajectories satisfying the desired conditions and demonstrating some generalization performance.
[1] Billard A, Calinon S, Dillmann R, et al. Robot programming by demonstration[EB/OL].[2018-12-29]. https://link.springer.com/content/pdf/10.1007/978-3-540-30301-5_60.pdf.
[2] Argall B D, Chernova S, Veloso M, et al. A survey of robot learning from demonstration[J]. Robotics and Autonomous Systems, 2009, 57(5):469-483.
[3] Pastor P, Hoffmann H, Asfour T, et al. Learning and generalization of motor skills by learning from demonstration[C]//IEEE International Conference on Robotics and Automation. Piscataway, USA:IEEE, 2009:763-768.
[4] Kober J, Mülling K, Krömer O, et al. Movement templates for learning of hitting and batting[C]//IEEE International Conference on Robotics and Automation. Piscataway, USA:IEEE, 2010:853-858.
[5] 姚峰,刘崇德,王玉甲,等.基于动态运动基元的轨迹学习方法[J].机器人,2018,40(4):560-568.Yao F, Liu C D, Wang Y J, et al. Trajectory learning based on dynamic movement primitives[J]. Robot, 2018, 40(4):560-568.
[6] Mülling K, Kober J, Kroemer O, et al. Learning to select and generalize striking movements in robot table tennis[J]. International Journal of Robotics Research, 2013, 32(3):263-279.
[7] Stulp F, Theodorou E A, Schaal S. Reinforcement learning with sequences of motion primitives for robust manipulation[J]. IEEE Transactions on Robotics, 2012, 28(6):1360-1370.
[8] Khansari-Zadeh S M, Billard A. Learning stable nonlinear dynamical systems with Gaussian mixture models[J]. IEEE Transactions on Robotics, 2011, 27(5):943-957.
[9] Hewitt A, Yang C G, Li Y, et al. DMP and GMR based teaching by demonstration for a KUKA LBR robot[C]//IEEE International Conference on Automation and Computing. Piscataway, USA:IEEE, 2017. DOI:10.23919/IConAC.2017.8081982.
[10] Gribovskaya E, Khansari-Zadeh S M, Billard A. Learning non-linear multivariate dynamics of motion in robotic manipulators[J]. International Journal of Robotics Research, 2011, 30(1):80-117.
[11] Chatzis S P, Korkinof D, Demiris Y. A quantum-statistical approach toward robot learning by demonstration[J]. IEEE Transactions on Robotics, 2012, 28(6):1371-1381.
[12] Calinon S, Li Z B, Alizadeh T, et al. Statistical dynamical systems for skills acquisition in humanoids[C]//IEEE-RAS International Conference on Humanoid Robots. Piscataway, USA:IEEE, 2012:323-329.
[13] Kim S, Gribovskaya E, Billard A. Learning motion dynamics to catch a moving object[C]//IEEE-RAS International Conference on Humanoid Robots. Piscataway, USA:IEEE, 2010:106-111.
[14] Koç O, Maeda G, Peters J. Online optimal trajectory generation for robot table tennis[J]. Robotics and Autonomous Systems, 2018, 105:121-137.
[15] Zhou F, de la Torre F. Canonical time warping for alignment of human behavior[C]//Advances in Neural Information Processing Systems. San Diego, USA:Neural Information Processing Systems Foundation, 2009:2286-2294.
[16] Zhou F, de la Torre F. Generalized canonical time warping[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(2):279-294.
[17] Yin X C, Chen Q J. Trajectory generation with spatio-temporal templates learned from demonstrations[J]. IEEE Transactions on Industrial Electronics, 2017, 64(4):3442-3451.
[18] Ramsay J O, Silverman B W. Functional data analysis[M]//In-ternational Encyclopedia of the Social & Behavioral Sciences. Oxford, UK:Pergamon, 2001:5822-5828.
[19] Hasan M A. On multi-set canonical correlation analysis[C]//In-ternational Joint Conference on Neural Networks. Piscataway, USA:IEEE, 2009:2640-2645.
[20] Zucker M, Ratliff N, Dragan A D, et al. CHOMP:Covariant Hamiltonian optimization for motion planning[J]. International Journal of Robotics Research, 2013, 32(9/10):1164-1193.
[21] 徐义春,万书振,解德祥.算法设计与分析[M].北京:清华大学出版社,2016.Xu Y C, Wan S Z, Xie D X. Algorithm design and analysis[M]. Beijing:Tsinghua University Press, 2016.