Learning Stable Nonlinear Dynamical Systems With Gaussian Mixture Models
Open accessThe DOI matches its Crossref record; 41 of the 43 references it cites were found in a registry or an index.
Abstract
Abstract—This paper presents a method for learning discrete robot motions from a set of demonstrations. We model a motion as a nonlinear autonomous (i.e. time-invariant) Dynamical System (DS), and define sufficient conditions to ensure global asymptotic stability at the target. We propose a learning method, called Stable Estimator of Dynamical Systems (SEDS), to learn the parameters of the DS to ensure that all motions follow closely the demonstrations while ultimately reaching in and stopping at the target. Time-invariance and global asymptotic stability at the target ensures that the system can respond immediately and appropriately to perturbations encountered during the motion. The method is evaluated through a set of robot experiments and on a library of human handwriting motions.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“In this model the convergence of all trajectories to the target is guaranteed, but the final velocity at the target is always zero (i.e. the target is an asymptotically stable point).”
From Learning to Play Minigolf: A Dynamical System-Based Approach (2012) · cites this paper as [2]
In the citing paper
“The resulting model from SEDS is globally asymptotically stable, i.e. starting from any point in the space, all trajectories converge to the origin.”
From Learning to Play Minigolf: A Dynamical System-Based Approach (2012) · cites this paper as [2]
In the citing paper
“The optimization constraints given by Eq. (10) ensure the global asymptotic stability of the function f̂(x;θ) (see [2]).”
From Learning to Play Minigolf: A Dynamical System-Based Approach (2012) · cites this paper as [2]
In the citing paper
“Gaussian distributions are used in PoMPs [114] and SEDs [115] for better generalisation behaviour.”
From Data-Driven Robotic Manipulation of Cloth-like Deformable Objects: The Present, Challenges and Future Prospects (2023) · cites this paper as [115]
In the citing paper
“We use the dynamical systems approach to encode motion from several demonstrations. This enables a rapid and reactive adaptation of the arm motion in the presence of sensor uncertainty.”
From Catching Objects in Flight (2014) · cites this paper as [20]
1 more claim about this paper
Create a free account to see every claim other papers make about this one.
Citations
Cites
A new look at the statistical model identification
Akaike, H. (1974) · IEEE Transactions on Automatic Control · doi:10.1109/tac.1974.1100705 · Cited by 51,879
Maximum Likelihood from Incomplete Data Via the EM Algorithm
Dempster, A. P., Laird, N. M., & Rubin, D. B. (1977) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.2517-6161.1977.tb01600.x · Cited by 50,444
Estimating the Dimension of a Model
Schwarz, G. (1978) · The Annals of Statistics · doi:10.1214/aos/1176344136 · Cited by 47,033
Slotine, J.-J. E., & Li, W. (1991) · CERN Document Server (European Organization for Nuclear Research) · Cited by 18,975
Gaussian Processes for Machine Learning
Rasmussen, C. E., & Williams, C. K. I. (2005) · doi:10.7551/mitpress/3206.001.0001 · Cited by 11,705
McLachlan, G., & Peel, D. (2000) · Wiley Series in Probability and Statistics · doi:10.1002/0471721182 · Cited by 10,991
Nonlinear programming: theory and algorithms
(1993) · Computers & Operations Research · doi:10.1016/0305-0548(93)90042-h · Cited by 5,779
ANew Look at the Statistical Model Identification
AKAlKE, H. (2011) · Cited by 4,177
Nonlinear Programming: Theory and Algorithms
Bazaraa, M. S., Sherali, H. D., & Shetty, C. M. (2006) · Wiley eBooks · doi:10.1002/0471787779 · Cited by 2,369
Sparse Gaussian Processes using Pseudo-inputs
Snelson, E., & Ghahramani, Z. (2005) · Cited by 1,415
Research works printed but not linked: 1
- schaal, 2002, Scalable locally weighted statistical techniques for real time robot learning, Appl Intell Spec Issue Scalable Robot Appl Neural Netw, 17, 49
Cited by
Items 301–400 of 772
Learning Complex Motion Plans using Neural ODEs with Safety and Stability Guarantees
Nawaz, F., Li, T., Matni, N., & Figueroa, N. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10611584 · Cited by 33
Reinhart, F., & Steil, J. J. (2014) · Autonomous Robots · doi:10.1007/s10514-014-9417-9 · Cited by 14
“The P (Ijspeert et al. 2003), NiVF (Lemme et al. 2013), etc.] GMM (Calinon et al. 2007; Mühli g et al. 2009), SEDS (Khansari-Zadeh and Billard 2011), Embedding ψ [SOM (Kohonen 1990), GTM (Bishop et al. 1998), MDS (Cox and Cox 2001), LLE (Sa u l and Roweis 2003), SNE (Hinton and Roweis 2002), etc.] multi-modal and cont…”
BackgroundMethodInfluentialTrajectory Generation With Spatio-Temporal Templates Learned From Demonstrations
Yin, X., & Chen, Q. (2016) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2016.2613843 · Cited by 14
“The GMR method can address the temporal and spatial disturbances while maintaining the shape of the generated trajectory [11].”
MethodNeural learning of stable dynamical systems based on extreme learning machine
Hu, J., Yang, Z., Wang, Z., Wu, X., & Ou, Y. (2015) · IEEE International Conference on Information and Automation (ICIA) · doi:10.1109/icinfa.2015.7279303 · Cited by 13
“where assuming f : R → R is a nonlinear continuous and continuously differentiable function with a single equilibrium point ẋ∗ = f(x∗) = 0, and ε is a zero mean gaussian noise [11].”
BackgroundInfluentialA learning from demonstration framework for implementation of a feeding task
Ettehadi, N., & Behal, A. (2018) · Encyclopedia with Semantic Computing and Robotic Intelligence · doi:10.1142/s2529737618500016 · Cited by 15
Learning and robotic imitation of therapist's motion and force for post-disability rehabilitation
Martinez, C., & Tavakoli, M. (2017) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc.2017.8122951 · Cited by 12
“A more detailed explanation about the algorithms can be found in [7], [9].”
BackgroundIncremental Learning Introspective Movement Primitives From Multimodal Unstructured Demonstrations
Wu, H., Xu, Z., Yan, W., Su, Q., Li, S., Cheng, T., & Zhou, X. (2019) · IEEE Access · doi:10.1109/access.2019.2947529 · Cited by 12
“As such, a set of multi-functional movement primitives [9] should be inspired for robots from human demonstrations rather than only learn the kinematic valuables [10], [11].”
BackgroundReactive motion generation on learned Riemannian manifolds
Beik-Mohammadi, H., Hauberg, S., Arvanitidis, G., Neumann, G., & Rozo, L. D. (2023) · The International Journal of Robotics Research · doi:10.1177/02783649231193046 · Cited by 16
“2020), or dynamical systems and probabilistic models (Ugur and Girgin 2020; Khansari-Zadeh and Billard 2011).”
BackgroundMethodPI-ELM: Reinforcement learning-based adaptable policy improvement for dynamical system
Hu, Y., Wang, X., Liu, Y., Ding, W., & Knoll, A. (2023) · Information Sciences · doi:10.1016/j.ins.2023.119700 · Cited by 12
A unified formulation of geometry-aware discrete dynamic movement primitives
Abu‐Dakka, F. J., Saveriano, M., & Kyrki, V. (2024) · Neurocomputing · doi:10.1016/j.neucom.2024.128056 · Cited by 12
Multifunctional physical reservoir computing in soft tensegrity robots
Terajima, R., Inoue, K., Nakajima, K., & Kuniyoshi, Y. (2025) · Chaos An Interdisciplinary Journal of Nonlinear Science · doi:10.1063/5.0273567 · Cited by 17
Hwang, J., Jung, M., Madapana, N., Kim, J.-H., Choi, M., & Tani, J. (2015) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2015.7363448 · Cited by 11
“Previous studies have used different methods such as Hidden Markov Model (HMM) [10], Gaussian Mixture Models (GMM) [11, 12] and Artificial Neural Networks (ANN) [13-15].”
MethodLearning robot anomaly recovery skills from multiple time-driven demonstrations
Wu, H., Yan, W., Xu, Z., Li, S., & Zhou, X. (2021) · Neurocomputing · doi:10.1016/j.neucom.2021.08.036 · Cited by 14
Generalization of orientational motion in unit quaternion space
Kramberger, A., Gams, A., Nemec, B., & Ude, A. (2016) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2016.7803366 · Cited by 12
“For example [14] used Gaussian mixture models.”
Salehi, I., Yao, G., & Dani, A. P. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8793891 · Cited by 12
“In [15], a nonlinear state space system is learned using a Stable Estimator of Dynamical Systems (SEDS) algorithm that preserves the property of a global convergence to an equilibrium point.”
BackgroundMethodConvex Nonparametric Formulation for Identification of Gradient Flows
Khosravi, M., & Smith, R. S. (2020) · IEEE Control Systems Letters · doi:10.1109/lcsys.2020.3000176 · Cited by 15
Figueroa, N., Faraji, S., Koptev, M., & Billard, A. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9197038 · Cited by 12
“1) Learning DS-based Walking Motion Planners: Works in DS-based learning from demonstration encode a nonlinear DS as a Linear Parameter Varying (LPV)-DS [29], [30], [31], [26], [32] that converges to an attractor x∗ as,”
BackgroundSelf-citationLearning to serve: an experimental study for a new learning from demonstrations framework
Koc, O., & Peters, J. (2019) · MPG.PuRe (Max Planck Society) · Cited by 12
Learning motion and impedance behaviors from human demonstrations
Saveriano, M., & Lee, D. (2014) · International Conference on Ubiquitous Robots and Ambient Intelligence · doi:10.1109/urai.2014.7057371 · Cited by 12
“Firstly, we underline some important concepts concerning the SEDS algorithm [2], used to train a motion primitive in the form of a globally asymptotically stable (GAS) DS.”
BackgroundMethodInfluentialGenerating a Style-Adaptive Trajectory from Multiple Demonstrations
Zhao, Y., Xiong, R., Li, F., & Dai, X. (2014) · International Journal of Advanced Robotic Systems · doi:10.5772/58723 · Cited by 12
“GMM has been used in different robotics applications, such as gesture imitation [9] and handwriting [10].”
MethodLearning Stabilizable Dynamical Systems via Control Contraction MetricsPreprint
Singh, S., Sindhwani, V., Slotine, J.-J. E., & Pavone, M. (2019) · EasyChair preprint · doi:10.29007/jz9w · Cited by 15
A survey on imitation learning for contact-rich tasks in robotics
Tsuji, T., Kato, Y., Solak, G., Zhang, H., Petrič, T., Nori, F., & Ajoudani, A. (2026) · The International Journal of Robotics Research · doi:10.1177/02783649261417694 · Cited by 76
“These works do not directly extend DMPs, however, they were historically developed in parallel and aimed to answer similar problems with similar principles (Khansari-Zadeh and Billard, 2011; Ureche et al., 2015).”
Manipulation Learning on Humanoid Robots
Gams, A., Petrič, T., Nemec, B., & Ude, A. (2022) · Current Robotics Reports · doi:10.1007/s43154-022-00082-9 · Cited by 15
“An important alternative is to build variability in the representation itself, such as with nonlinear dynamic systems [24] and probabilistic movement primitives [72].”
BackgroundA geometric optimal control approach for imitation and generalization of manipulation skills
Ti, B., Razmjoo, A., Gao, Y., Zhao, J., & Calinon, S. (2023) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2023.104413 · Cited by 11
Hu, J., Liu, J., Guo, Y., Cao, Z., Chen, X., & Zhang, C. (2023) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2023.107040 · Cited by 12
Motion Planning and Inertia-Based Control for Impact Aware Manipulation
Khurana, H., & Billard, A. (2023) · IEEE Transactions on Robotics · doi:10.1109/tro.2023.3319895 · Cited by 11
“They are heavily used in learning from demonstration [24], where the data from demonstrated trajectories is used tolearnstableDSs.”
BackgroundSelf-citationRuan, S., Liu, W., Wang, X., Meng, X., & Chirikjian, G. S. (2024) · IEEE Transactions on Robotics · doi:10.1109/tro.2024.3390052 · Cited by 16
“The data for LfD benchmarks include three types, which are as follows: 1) LASA hand-writing dataset [63]; 2) end effector positions generated in simulation; 3) real-world kinesthetic teaching for several common daily tasks.”
MethodInstructing Robots by Sketching: Learning from Demonstration via Probabilistic Diagrammatic Teaching
Zhi, W., Zhang, T., & Johnson‐Roberson, M. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10611349 · Cited by 27
“These include Dynamical Movement Primitive (DMP) approaches [4], [5], Probabilistic Movement Primitives [6], and stable dynamical system approaches [7], [8], [9].”
BackgroundLearning Stable State-Dependent Variable Impedance Control for Compliant Manipulation
Jin, Z., Liu, A., Zhang, W., Yu, L. J., & Yang, C. (2024) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2024.3466950 · Cited by 11
“In [16], a modified PI 2 approach was proposed where the DMP framework is replaced by the SEDS [17] framework, which allows the obtained variable impedance law to be fully state-dependent and hence improves the adaptability.”
Combining federated learning and control: A survey
Weber, J., Gurtner, M., Lobe, A., Trachte, A., & Kugi, A. (2024) · IET Control Theory and Applications · doi:10.1049/cth2.12761 · Cited by 23
“In [128], the combination of SINDY - sparse identification of nonlinear dynamics, see [129] - and MPC is proposed to enhance the control performance.”
BackgroundRana, M. A., Chen, D., Williams, J. M., Chu, V. C.-M., Ahmadzadeh, S. R., & Chernova, S. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9197470 · Cited by 18
“Based on the choice of motion generation strategy, the majority of skill learning approaches can be further broadly categorized into four groups: time-invariant (stable) dynamical systems [2], [3], [13], time-dependent dynamical systems [7], [9], [10], time-dependent statistical methods [8], [14], [15], and arclength…”
BackgroundLearning and extrapolation of robotic skills using task-parameterized equation learner networks
Pérez-Villeda, H. M., Piater, J., & Saveriano, M. (2022) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2022.104309 · Cited by 17
“Stable Dynamical Systems (DSs) are capable of planning converging motions from any point of the state space and have been effectively used as MP [10, 11, 12, 13].”
MethodJansson, D., & Medvedev, A. V. (2014) · European Control Conference (ECC) · doi:10.1109/ecc.2014.6862207 · Cited by 11
Representing movement primitives as implicit dynamical systems learned from multiple demonstrations
Krug, R., & Dimitrovz, D. (2013) · International Conference on Advanced Robotics (ICAR) · doi:10.1109/icar.2013.6766505 · Cited by 11
“However, as stated by the authors in [11], with increasing number of DoF the learning problem can become intractable.”
BackgroundLearning Accurate and Stable Dynamical System Under Manifold Immersion and Submersion
Jin, S., Wang, Z., Ou, Y., & Feng, W. (2019) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2019.2892207 · Cited by 11
“In spite of its competence over the SEDS in learning a larger set of complex motions, the first procedure of learning a satisfactory Lyapunov function and the second procedure of estimating the demonstration dynamics are two separate processes.”
BackgroundMethodInfluentialDuan, J., Ou, Y., Xu, S., & Liu, M. (2019) · Neurocomputing · doi:10.1016/j.neucom.2019.07.081 · Cited by 11
“Sequential Learning of the Kinematic Profile Various techniques can be applied to capture the kinematics of the task, such as DMP [4], Dynamical System (DS) [6, 7] and so on.”
MethodLearning Dynamics Models with Stable Invariant Sets
Takeishi, N., & Kawahara, Y. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i11.17176 · Cited by 12
Persistent Homology for Learning Densities with Bounded Support
Pokorny, F. T., Ek, C. H., Hedvig, E., & Kragic, K. D. (2012) · Cited by 11
Gaussian process for 6-DoF rigid motions
Lang, M., Kleinsteuber, M., & Hirche, S. (2017) · Autonomous Robots · doi:10.1007/s10514-017-9683-4 · Cited by 12
Learning Generalizable Robot Skills from Demonstrations in Cluttered Environments
Rana, M. A., Mukadam, M., Ahmadzadeh, S. R., Chernova, S., & Boots, B. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8593624 · Cited by 12
A Robot Skill Learning Framework Based on Compliant Movement Primitives
Dou, S., Xiao, J., Zhao, W., Yuan, H., & Liu, H. (2022) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-022-01605-4 · Cited by 10
“In [16], GMM was used to learn discrete robot motion from multiple demonstrations.”
MethodModel gradient: unified model and policy learning in model-based reinforcement learning
Jia, C., Zhang, F., Xu, T., Pang, J.-C., Zhang, Z., & Yu, Y. (2023) · Frontiers of Computer Science · doi:10.1007/s11704-023-3150-5 · Cited by 13
Probabilistic movement primitives based multi-task learning framework
Yue, C., Gao, T., Lu, L., Lin, T., & Wu, Y. (2024) · Computers & Industrial Engineering · doi:10.1016/j.cie.2024.110144 · Cited by 10
Feature Space Decomposition for effective robot adaptation
Zhang, C., Zhang, H., & Parker, L. E. (2015) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2015.7353410 · Cited by 9
“Khansari-Zadeh and Billard [12] modulated the dynamical system using trajectories encoded by a GMM.”
MethodLearning Deep Robotic Skills on Riemannian Manifolds
Wang, W., Saveriano, M., & Abu‐Dakka, F. J. (2022) · IEEE Access · doi:10.1109/access.2022.3217800 · Cited by 11
“The dataset we used is based on LASA dataset [26] which consists of 30 shapes, each with 7 trajectories of 1000 points in 2- dimensional.”
BackgroundMethodOn the Benefits of Learning to Route in Mixture-of-Experts Models
Dikkala, N., Ghosh, N., Meka, R., Panigrahy, R., Vyas, N., & Wang, X. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.583 · Cited by 50
“We choose mixture of Gaussians as it is known to be a highly powerful modeling framework in practice (Khansari-Zadeh and Billard, 2011; Varadara-jan et al., 2013) while also theoretically capable of representing arbitrary distributions using Gaussians as basis functions.”
Behavior sequencing based on demonstrations: a case of a humanoid opening a door while walking
González-Fierro, M., García, D. H., Nanayakkara, T., & Balaguer, C. (2015) · Advanced Robotics · doi:10.1080/01691864.2014.992955 · Cited by 12
“There are many studies conducted in the area of Learning from Demonstration using GMM and GMR to encode kinesthetic trajectories and generalize them to perform a robot movement [8–10 , Our work takes some ideas from [9] where a humanoid robot use LfD to initially learn a pick and place task.”
BackgroundSynthesizing compliant reaching movements by searching a database of example trajectories
Deniša, M., Petrič, T., Asfour, T., & Ude, A. (2013) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2013.7030026 · Cited by 12
“A set of movement trajectories, which partly share a similar course of movement, can be used to discover new movements.”
BackgroundMotion reproduction using time-scaling for adaptation to difference in environmental location
Nozaki, T., Mizoguchi, T., & Ohnishi, K. (2014) · IEEE International Conference on Industrial Technology (ICIT) · doi:10.1109/icit.2014.6894970 · Cited by 10
Learning motion primitives from demonstration
Chi, M., Yao, Y., Liu, Y., Teng, Y., & Zhong, M. (2017) · Advances in Mechanical Engineering · doi:10.1177/1687814017737260 · Cited by 10
A learning framework for semantic reach-to-grasp tasks integrating machine learning and optimization
Deng, Z., Zheng, X., Zhang, L., & Zhang, J. (2018) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2018.08.001 · Cited by 12
Property-Aware Robot Object Manipulation: a Generative Approach
Garello, L., Lastrico, L., Rea, F., Mastrogiovanni, F., Noceti, N., & Sciutti, A. (2021) · IEEE International Conference on Development and Learning (ICDL) · doi:10.1109/icdl49984.2021.9515667 · Cited by 12
“[9], who focused on the Careful (C) / Not Careful (NC) manipulation of objects, and tried to model the two movement classes using Gaussian Mixture Models (GMMs) [10].”
Contraction-Based Methods for Stable Identification and Robust Machine Learning: a Tutorial
Manchester, I. R., Revay, M., & Wang, R. (2021) · IEEE Conference on Decision and Control (CDC) · doi:10.1109/cdc45484.2021.9683128 · Cited by 12
“for polynomial models [3], [4], [5], Gaussian mixture models [22], and recurrent neural networks [23], [8], [10], however the problem is substantially more complex than the linear case as there are many different definitions of nonlinear stability and even verification of stability of a given model is challenging.”
BackgroundAlibeigi, M., Ahmadabadi, M. N., & Araabi, B. N. (2016) · IEEE Transactions on Robotics · doi:10.1109/tro.2016.2623817 · Cited by 12
“In this method, all perceptual concepts are incrementally learned in a single recurrent neural network through the proposed memory rehearsal.”
MethodComplex Sequential Tasks Learning with Bayesian Inference and Gaussian Mixture Model
Zhang, H., Han, X., Zhang, W., & Zhou, W. (2018) · IEEE International Conference on Robotics and Biomimetics (ROBIO) · doi:10.1109/robio.2018.8665229 · Cited by 10
“Actually by optimizing the parameters of the joint distribution of (ξ, ξ̇) to make A and b meet some specific conditions [38], the encoded model in (13) can ensure global convergence.”
MethodOn Gaussian Process Based Koopman Operators
Lian, Y., & Jones, C. N. (2020) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2020.12.217 · Cited by 10
DyNODE: Neural Ordinary Differential Equations for Dynamics Modeling in Continuous ControlPreprint
Alvarez, V. M. M., Rosca, R., & Falcutescu, C. G. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2009.04278 · Cited by 39
“, 2015b; Montgomery and Levine, 2016) or a mixture of Gaussian processes (Khansari-Zadeh and Billard, 2011).”
BackgroundA Review of Symbolic, Subsymbolic and Hybrid Methods for Sequential Decision Making
Núñez-Molina, C., Mesejo, P., & Fernández-Olivares, J. (2024) · ACM Computing Surveys · doi:10.1145/3663366 · Cited by 14
“Two possible solutions are to approximate the entire next state distribution [93] and to learn a generative model from which we can draw samples [36].”
Learning and correcting robot trajectory keypoints from a single demonstration
Iturrate, I., Ostergaard, E. H., Rytter, M., & Savarimuthu, T. R. (2017) · International Conference on Control, Automation and Robotics (ICCAR) · doi:10.1109/iccar.2017.7942660 · Cited by 10
Jin, Z., Qin, D., Liu, A., Zhang, W., & Li, Y. (2023) · Robotics and Computer-Integrated Manufacturing · doi:10.1016/j.rcim.2023.102526 · Cited by 12
EC2: Emergent Communication for Embodied Control
Mu, Y., Yao, S., Ding, M., Luo, P., & Gan, C. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00648 · Cited by 23
Kinesthetic teaching via Fast Marching Square
Gómez, J. V., Álvarez, D., Garrido, S., & Moreno, L. E. (2012) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2012.6385497 · Cited by 9
“point-to-point trajectories) [7], where the robot performs a movement keeping the motion as similar as possible to the demonstrations.”
BackgroundTrajectory learning from human demonstrations via manifold mapping
Hiratsuka, M., Makondo, N., Rosman, B., & Hasegawa, O. (2016) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2016.7759579 · Cited by 12
“Reward function learning [10], statistical modeling [11] and dynamical systems [12-15] are amongst the most popular approaches.”
BackgroundAdaptive motion planning framework by learning from demonstration
Li, X., Cheng, H., & Liang, X. (2019) · Industrial Robot the international journal of robotics research and application · doi:10.1108/ir-10-2018-0216 · Cited by 9
Li, C., Li, Z., Jiang, Z., Cui, S., Liu, H., & Cai, H. (2021) · Science China Technological Sciences · doi:10.1007/s11431-021-1901-x · Cited by 9
“Step 1: We build the general skill model with the dynamic system theory [20], as shown in eq.”
MethodCalinon, S., & Lee, D. (2018) · Springer eBooks · doi:10.1007/978-94-007-7194-9_68-2 · Cited by 9
Hierarchical Policy Blending as Inference for Reactive Robot Control
Hansel, K., Urain, J., Peters, J., & Chalvatzaki, G. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10161374 · Cited by 18
“Motion primitives provide learning-based generation of reactive, stable behavior [4], [10], [48].”
BackgroundDeniša, M., & Ude, A. (2015) · International Journal of Advanced Robotic Systems · doi:10.5772/61036 · Cited by 9
“, (xm, Tm)} as a dynamic movement primitive (DMP) [10].”
BackgroundMethodLearning cost function and trajectory for robotic writing motion
Yin, H., Paiva, A., & Billard, A. (2014) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2014.7041425 · Cited by 9
“The letter trajectories are collected from dataset [13].”
MethodSelf-citationDynamical system based variable admittance control for physical human-robot interaction
Bian, F., Ren, D., Li, R., Liang, P., Wang, K., & Zhao, L. (2020) · Industrial Robot the international journal of robotics research and application · doi:10.1108/ir-12-2019-0258 · Cited by 9
Neural Liénard system: learning periodic manipulation skills through dynamical systems
Zhang, H., Cheng, L., Zhang, Y., & Wang, Y. (2024) · Science China Information Sciences · doi:10.1007/s11432-023-4219-8 · Cited by 8
Stable-BC: Controlling Covariate Shift With Stable Behavior Cloning
Mehta, S. A., Ciftci, Y. U., Ramachandran, B., Bansal, S., & Losey, D. P. (2025) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2025.3526439 · Cited by 39
A Lyapunov-Based Framework for Trajectory Planning of Wheeled Vehicle Using Imitation Learning
Lai, J., Wu, Z., Ren, Z., Chen, C., Tan, Q., & Xie, S. (2025) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2025.3541409 · Cited by 8
“The resulting estimated model ˆ f ( σ ) often contains numerous erroneous attractors [41].”
InfluentialCompliant Motion Adaptation with Dynamical System during Robot-Environment Interaction
Huang, H., Yang, C., & Su, C. (2020) · IEEE/ASME International Conference on Advanced Intelligent Mechatronics · doi:10.1109/aim43001.2020.9158894 · Cited by 8
“Motion adaptive performance of the proposed scheme is first evaluated by controlling manipulator to move in the free and constraint space against a set of DS motions which are generated from [20].”
MethodBayesian Disturbance Injection: Robust Imitation Learning of Flexible Policies
Oh, H., Sasaki, H., Michael, B., & Matsubara, T. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561573 · Cited by 10
Robot Learning from Demonstration Using Elastic Maps
Hertel, B., Pelland, M., & Ahmadzadeh, S. R. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981241 · Cited by 11
“3 where different downsampling methods with M = 1000 and N = 8 are used on a 2D trajectory from the LASA dataset [20].”
MethodInfluentialJaquier, N., Rozo, L. D., & Asfour, T. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10611701 · Cited by 8
“Most prior works assume the state x to evolve in an Euclidean space [33]–[38].”
BackgroundMethodMMP++: Motion Manifold Primitives With Parametric Curve Models
Lee, Y. (2024) · IEEE Transactions on Robotics · doi:10.1109/tro.2024.3444068 · Cited by 16
“1) Dynamical system-based approaches, such as dynamic movement primitives [8], [9], [13], [14], [15], [16] and stable dynamical systems [17], [18], [19], [20], [21], [22].”
BackgroundEnayati, N., Mariani, S., Wahrburg, A., & Zanchettin, A. M. (2020) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2020.12.2687 · Cited by 8
Towards Data-driven LQR with Koopmanizing Flows⋆
Bevanda, P., Beier, M., Heshmati-Alamdari, S., Sosnowski, S., & Hirche, S. (2022) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2022.07.601 · Cited by 8
NetLfD: Network-Aware Learning From Demonstration for In-Contact Skills via Teleoperation
Güleçyüz, B., von Büren, V., Xu, X. W., & Steinbach, E. G. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3313494 · Cited by 8
Hwang, C., Chen, B., Huang, H.-H., & Syu, H.-T. (2014) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2014.12.003 · Cited by 8
“KhansariZadeh and Billard [5] decomposed the motions into point-to-point for robots to imitate.”
BackgroundEncoding bi-manual coordination patterns from human demonstrations
Ureche, A. L. P., & Billard, A. (2014) · ACM/IEEE International Conference on Human-Robot Interaction (HRI) · doi:10.1145/2559636.2559844 · Cited by 8
“The arm motion in each segment is encoded as a non-linear dynamical system [4].”
BackgroundSelf-citationLearning periodic motions from human demonstrations using transverse contraction analysis
Ravichandar, H. C., Thota, P. K., & Dani, A. P. (2016) · Proceedings of the ... American Control Conference/Proceedings of the American Control Conference · doi:10.1109/acc.2016.7526121 · Cited by 9
“The work most related to this paper is learning from demonstration (LfD) using Gaussian Mixture Model (GMM) proposed in [2], [15].”
MethodGaussian process based model predictive controller for imitation learning
Joukov, V., & Kulić, D. (2017) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2017.8246971 · Cited by 8
“Next, in simulation, we use the Barret WAM manipulator and the Lasa hand writing dataset [9] to show how the proposed approach can be used with GPs learned from real human motion.”
BackgroundMethodOne-Shot Multi-Path Planning for Robotic Applications Using Fully Convolutional Networks
Kulvičius, T., Herzog, S., Lüddecke, T., Tamošiūnaitė, M., & Wörgötter, F. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9196719 · Cited by 9
“In real applications, the points of the path P (x, y) can then be used as via points to generate trajectories using conventional methods such as splines [6], [27] or more advanced state-of-the-art methods such as dynamic movement primitives (DMPs, [13]), Gaussian mixture models (GMMs, [16]), probabilistic movement pri…”
MethodContinuous Modeling of Affordances in a Symbolic Knowledge Base
Bozcuoğlu, A. K., Furuta, Y., Okada, K., Beetz, M., & Inaba, M. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8968220 · Cited by 8
“There are some related works on Gaussian Mixture Models in robot learning applications such as learning by demonstration [20], [21], [22] and model-based control [23].”
BackgroundAlmost Surely Stable Deep DynamicsPreprint
Lawrence, N. P., Loewen, P. D., Forbes, M. G., Backström, J. U., & Gopaluni, R. B. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.14722 · Cited by 10
“Khansari-Zadeh and Billard [25] consider Gaussian mixture models for learning continuous-time dynamical systems but only enforce stability of the means.”
BackgroundData-driven learning and control of nonlinear system dynamics
Becerra, Y., & Acosta, J. Á. (2024) · Nonlinear Dynamics · doi:10.1007/s11071-024-10149-1 · Cited by 7
de Carvalho, G. P., Sawanobori, T., & Horii, T. (2025) · IEEE Access · doi:10.1109/access.2025.3552225 · Cited by 12
A Non-Linear Manifold Alignment Approach to Robot Learning from Demonstrations
Makondo, N., Hiratsuka, M., Rosman, B., & Hasegawa, O. (2018) · Journal of Robotics and Mechatronics · doi:10.20965/jrm.2018.p0265 · Cited by 7
“Statistical modeling [5, 8] and dynamical systems [9, 10] are amongst the most popular approaches.”
BackgroundLearning robot reaching motions by demonstration using nonlinear autoregressive models
dos Santos, R. F., Pereira, G. A. S., & Aguirre, L. A. (2018) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2018.06.006 · Cited by 7
“Another well-known method that generate autonomous DS models is the Stable Estimator of Dynamical Systems (SEDS) [4].”
BackgroundMethodResultInfluentialWu, H., Zhai, D., & Xia, Y. (2024) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2024.3384602 · Cited by 8
“In the proof of the convergence of the imitation learning method, the methods that need to be proved include DMP and the stable estimator of dynamical systems (SEDS) [26] method.”
Zhang, Y., Zou, Y., Li, H., Zhang, H., & Cheng, L. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10610633 · Cited by 9
“The first work that combines Lyapunov theorems with DS learning is known as the Stable Estimator of Dynamical Systems (SEDS) [4].”
BackgroundJadav, S., Heidersberger, J., Ott, C., & Lee, D. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10610761 · Cited by 7
“The Stable Estimator of Dynamical System (SEDS) is an algorithm based on Gaussian mixture models, offering stability guarantees by attracting towards a global point [7].”
BackgroundMethodPUMA: Deep Metric Imitation Learning for Stable Motion Primitives
Pérez‐Dattari, R., Santina, C. D., & Kober, J. (2024) · Advanced Intelligent Systems · doi:10.1002/aisy.202400144 · Cited by 8
“Nevertheless, previous methods for stable motion generation have been initially designed under the assumption that the state space is Euclidean [4, 2, 3, 9].”
BackgroundMethodInfluentialTeleoperator Imitation with Continuous-Time Safety
Khadir, B. E., Varley, J., & Sindhwani, V. (2019) · Robotics: Science and Systems · doi:10.15607/rss.2019.xv.038 · Cited by 9
Lin, H., Gao, Q., Chu, X., Dou, Q., Deguet, A., Kazanzides, P., & Au, K. W. S. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2021.3062351 · Cited by 11
“Particularly, Linear Regression [7], [8], Gaussian Mixture Regression [9], Gaussian Process Regression [10], Support Vector Machine [11], Feedforward Neural Network [12]–[14] and Recurrent Neural Network [15], [16] were applied to approximate the system dynamics of robots based on measured data.”
MethodReview of the techniques used in motor‐cognitive human‐robot skill transfer
Guan, Y. H., Wang, N., & Yang, C. (2021) · Cognitive Computation and Systems · doi:10.1049/ccs2.12025 · Cited by 8
“Stable estimator of dynamic systems (SEDS) is complementary to DMPs [69, 70].”
BackgroundMethodSemi-Autonomous Cooperative Tasks in a Multi-Arm Robotic Surgical Domain
Deniša, M., Schwaner, K. L., Iturrate, I., & Savarimuthu, T. R. (2021) · International Conference on Advanced Robotics (ICAR) · doi:10.1109/icar53236.2021.9659445 · Cited by 11
“Gaussian mixture models [12], [13], Probabilistic Movement Primitives [14] or Hidden Markov Models (HMMs) [15] can also be used to encode multiple demonstrations.”
Background
Metadata
- Authors
- Khansari-Zadeh, S. M., & Billard, A.
- Affiliations
- École Polytechnique Fédérale de Lausanne (CH)
- ORCID
- Billard, Aude
- Venue
- IEEE Transactions on Robotics
- Published
- 2011-10
- Volume
- 27
- Issue
- 5
- Pages
- 943-957
- ISSN
- 1552-3098 (ISSN-L)
- 1941-0468
- OpenAlex
- W2129202194
- Cited by
- 771
- References
- 43
- Citations by year
- 2026: 41 · 2025: 71 · 2024: 73 · 2023: 74 · 2022: 47 · 2021: 67 · 2020: 65 · 2019: 69 · 2018: 52 · 2017: 43 · 2016: 31 · 2015: 42 · 2014: 41 · 2013: 28 · 2012: 22
- FWCI
- 27.98
- Type
- journal article
- Language
- en
- Open access
- green
- infoscience.epfl.ch
- Locations
- IEEE Transactions on Robotics (publishedVersion)
- Infoscience (Ecole Polytechnique Fédérale de Lausanne) (submittedVersion)
- Infoscience (Ecole Polytechnique Fédérale de Lausanne) (submittedVersion, cc-by-nc-nd)
- Funders
- European Commission (FP7 ICT, 248311)
- FP7 Information and Communication Technologies
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Physical Sciences › Engineering › Control and Systems Engineering › Robot Manipulation and Learning
- Life Sciences › Neuroscience › Cognitive Neuroscience › Motor Control and Adaptation
- Keywords
- learning dynamical systems, global asymptotic stability, Gaussian mixture model, time-invariant systems, robustness to disturbances
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.