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 201–300 of 772
Learning to control planar hitting motions in a minigolf-like task
Kronander, K., Khansari-Zadeh, M. S. M., & Billard, A. (2011) · 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems · doi:10.1109/iros.2011.6094402 · Cited by 23
“In [11] it is shown how demonstrations, assumed instances of a stable DS:”
BackgroundSelf-citationLuo, X. L., Xiao, M., Ding, Y., & Ding, H. (2020) · Sensors and Actuators A Physical · doi:10.1016/j.sna.2020.112227 · Cited by 22
Probabilistic Motion Prediction and Skill Learning for Human-to-Cobot Dual-Arm Handover Control
Yan, Z., He, W., Wang, Y., Sun, L., & Yu, X. (2022) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2022.3182973 · Cited by 21
“GMR can compute full covariance matrices encoding the correlation between high-dimensional state variables and predict reasonable state variables for outlier state variables based on current models [13].”
BackgroundA Learning-Based Object Tracking Strategy Using Visual Sensors and Intelligent Robot Arm
Xu, S., Chen, K., Ou, Y., Wang, Z., & Yang, C. (2022) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2022.3213730 · Cited by 25
Task-unit based trajectory generation for excavators utilizing expert operator skills
Feng, C., Shen, Y., Wang, J., Wang, Q., Suo, Z., & Fa, S. (2023) · Automation in Construction · doi:10.1016/j.autcon.2023.105247 · Cited by 22
Neurally Imprinted Stable Vector Fields
Lemme, A., Neumann, K., Reinhart, F., & Steil, J. J. (2013) · PUB – Publications at Bielefeld University (Bielefeld University) · Cited by 22
Learning manipulation actions from human demonstrations
Welschehold, T., Dornhege, C., & Burgard, W. (2016) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2016.7759555 · Cited by 22
“Alternative approaches for time independent motion learning could also be used [20].”
MethodMurata, S., Li, Y., Arie, H., Ogata, T., & Sugano, S. (2018) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2018.2797260 · Cited by 22
“As an alternative approach, Khansari-Zadeh and Billard [8] proposed a dynamical systems framework called a stable estimator of dynamical systems.”
MethodA Passivity-Based Strategy for Coaching in Human-Robot Interaction
Landi, C. T., Ferraguti, F., Fantuzzi, C., & Secchi, C. (2018) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2018.8460836 · Cited by 22
“In [5], [6] the authors propose a Stable Estimator of Dynamical Systems to model the motion”
BackgroundAn image-based trajectory planning approach for robust robot programming by demonstration
Vakanski, A., Janabi‐Sharifi, F., & Mantegh, I. (2017) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2017.09.012 · Cited by 21
Learning Stable Vector Fields on Lie Groups
Urain, J., Tateo, D., & Peters, J. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3219019 · Cited by 31
“During the last two decades, there has been vast research on learning policy architectures [3], [4], [5], [6], [7] that guarantee a set of desirable inductive biases.”
BackgroundMethodInfluentialLiang, C., Kamat, V. R., Menassa, C. C., & McGee, W. (2021) · Journal of Computing in Civil Engineering · doi:10.1061/(asce)cp.1943-5487.0001004 · Cited by 25
Type-2 Fuzzy Model-Based Movement Primitives for Imitation Learning
Sun, D. M., Liao, Q., & Loutfi, A. (2022) · IEEE Transactions on Robotics · doi:10.1109/tro.2022.3152685 · Cited by 23
Towards Learning of Generic Skills for Robotic Manipulation
Metzen, J. H., Fabisch, A., Senger, L., Fernández, J. D. G., & Kirchner, E. A. (2013) · Künstliche Intell. · doi:10.1007/s13218-013-0280-1 · Cited by 23
“, SEDS [20], since the optimal weights for a given demonstration can be determined by a closed-form formula.”
BackgroundPervez, A., Latifee, H. O., Ryu, J.-H., & Lee, D. (2019) · Autonomous Robots · doi:10.1007/s10514-019-09853-4 · Cited by 27
A passive robot controller aiding human coaching for kinematic behavior modifications
Papageorgiou, D., Kastritsi, T., & Doulgeri, Z. (2019) · Robotics and Computer-Integrated Manufacturing · doi:10.1016/j.rcim.2019.101824 · Cited by 21
Reinforcement learning of motor skills using Policy Search and human corrective advice
Celemin, C. E., Maeda, G. J., Ruíz-del-Solar, J., Peters, J., & Kober, J. (2019) · The International Journal of Robotics Research · doi:10.1177/0278364919871998 · Cited by 31
“This work is focused on movement primitives, such as DMPs or ProMPs, whose evolutions are defined by a phase variable zt, but other representations are also possible (Khansari-Zadeh and Billard, 2011).”
BackgroundInfluentialLearning and Sequencing of Object-Centric Manipulation Skills for Industrial Tasks
Rozo, L. D., Guo, M., Kupcsik, A., Todescato, M., Schillinger, P., Giftthaler, M., Ochs, M., Spies, M., Waniek, N., Kesper, P., et al. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9341570 · Cited by 31
Task-Based Role Adaptation for Human-Robot Cooperative Object Handling
Ansari, R. J., & Karayiannidis, Y. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2021.3064498 · Cited by 21
“In contrast to global dynamical systems defined based on attractive equilibrium points [11], [13], here we propose to employ three generic local dynamical systems with distinct geometries: free motion, rotation, and translation of an object.”
BackgroundMethodInfluentialA probabilistic approach to robot trajectory generation
Paraschos, A., Neumann, G. M., & Peters, J. (2013) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2013.7030017 · Cited by 25
“While most of these properties are implemented by one or more MP representations [1], [2], [8], a unified framework is missing that combines all these properties in a principled way without relying on heuristics.”
BackgroundLukic, L., Santos-Victor, J., & Billard, A. (2014) · Biological Cybernetics · doi:10.1007/s00422-014-0591-9 · Cited by 23
“We use the stable estimator of dynamical systems (SEDS) (Khansari-Zadeh and Billard 2011) to compute the GMM parameters.”
MethodInfluentialSelf-citationRobust learning from demonstrations using multidimensional SAX
Mohammad, Y. F. O., & Nishida, T. (2014) · International Conference on Control, Automation and Systems (ICCAS) · doi:10.1109/iccas.2014.6987960 · Cited by 20
Incremental Skill Learning of Stable Dynamical Systems
Saveriano, M., & Lee, D. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8594474 · Cited by 24
“The stable estimator of dynamical systems (SEDS) in [10] generates stable motions from a non-linear DS, represented by GMM.”
BackgroundMethodAn Energy-based Approach to Ensure the Stability of Learned Dynamical Systems
Saveriano, M. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9196978 · Cited by 20
“pabilities outside the demonstration area [5]. An alternative approach is to encode the motion into a state-dependent DS. This idea is exploited by the stable estimator of dynamical systems (SEDS) in [3], where the parameters of a Gaussian mixture regression (GMR) are constrained to ensure global stability. However, S…”
BackgroundMethodInfluentialModel predictive optimization for imitation learning from demonstrations
Hu, Y., Cui, M., Duan, J., Liu, W., Huang, D., Knoll, A., & Chen, G. (2023) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2023.104381 · Cited by 20
Learning stable robotic skills on Riemannian manifolds
Saveriano, M., Abu‐Dakka, F. J., & Kyrki, V. (2023) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2023.104510 · Cited by 24
“1 The Stable Estimator of Dynamical Systems (SEDS) [6] is one of the first approaches that learn stable and autonomous DSs.”
BackgroundMethodHuman intention inference through interacting multiple model filtering
Ravichandar, H. C., & Dani, A. P. (2015) · IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI) · doi:10.1109/mfi.2015.7295812 · Cited by 25
“Learning the dynamics of the human motion and giving robot an ability to infer the goal locations of a human co-worker are important problems in the context of safe and efficient collaboration [3], [4].”
BackgroundLearning of Parameters in Behavior Trees for Movement Skills
Mayr, M., Chatzilygeroudis, K. I., Ahmad, F., Nardi, L., & Krueger, V. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636292 · Cited by 29
“toward perturbations and stability guarantees [3], [7], [4], [5], which are desirable properties of a robotic system.”
Gams, A., Petrič, T., Nemec, B., & Ude, A. (2014) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2014.7041354 · Cited by 20
“Other trajectory representations include splines [4], mixture models [5] and different dynamical systems, which can also be adapted based on force feedback [6].”
BackgroundOrientation Probabilistic Movement Primitives on Riemannian ManifoldsPreprint
Rozo, L. D., & Dave, V. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.15036 · Cited by 20
“Probabilistic movement primitives (ProMPs) [4], Stable Estimators of Dynamical Systems (SEDS) [8], Task-Parameterized Gaussian Mixture Models (TP-GMMs) [5], Kernelized MPs (KMPs) [9] and Conditional Neural MPs (CNMPs) [6] are some of the recent probabilistic models to represent motion primitives (MPs).”
BackgroundRobot learning from multiple demonstrations with dynamic movement primitive
Chen, C., Yang, C., Zeng, C., Wang, N., & Li, Z. (2017) · International Conference on Advanced Robotics and Mechatronics (ICARM) · doi:10.1109/icarm.2017.8273217 · Cited by 24
“Stable estimator of dynamical systems (SEDS) is another approach of PbD with dynamic system representation [10].”
BackgroundLearning Reactive Motion Policies in Multiple Task Spaces from Human Demonstrations.
Rana, M. A., Li, A., Ravichandar, H. C., Mukadam, M., Chernova, S., Fox, D., Boots, B., & Ratliff, N. (2019) · Conference on Robot Learning · Cited by 19
Learning dynamical systems with bifurcations
Khadivar, F., Lauzana, I., & Billard, A. (2020) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2020.103700 · Cited by 19
Self-citationReward Learning From Very Few Demonstrations
Eteke, C., Kebüde, D., & Akgün, B. (2020) · IEEE Transactions on Robotics · doi:10.1109/tro.2020.3038698 · Cited by 25
“The ideas can also be used together as done with stable estimator of dynamical Systems [15].”
BackgroundDesign of Hesitation Gestures for Nonverbal Human-Robot Negotiation of Conflicts
Moon, A., Hashmi, M., der Loos, H. F. M. V., Croft, E. A., & Billard, A. (2021) · ACM Transactions on Human-Robot Interaction · doi:10.1145/3418302 · Cited by 22
Self-citationRobot Skill Adaptation via Soft Actor-Critic Gaussian Mixture Models
Nematollahi, I., Rosete-Beas, E., Rpfer, A., Welschehold, T., Valada, A., & Burgard, W. (2022) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra46639.2022.9811770 · Cited by 32
ADAM: a robotic companion for enhanced quality of life in aging populations
Mora, A., Prados, A., Méndez, A., Espinoza, G., González, P., Lopez, B., Muñoz, V., Moreno, L. E., Garrido, S., & Barber, R. (2024) · Frontiers in Neurorobotics · doi:10.3389/fnbot.2024.1337608 · Cited by 25
Colomé, A., & Torras, C. (2014) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2014.6942742 · Cited by 23
“Our three proposals have been experimentally tested and the obtained results show that similar (or even better) performance can be obtained at a significantly lower computational cost by reducing the dimensionality of the exploration space.”
BackgroundLearning Stable Stochastic Nonlinear Dynamical Systems
Umlauft, J., & Hirche, S. (2017) · International Conference on Machine Learning · Cited by 19
Lechner, M., Hasani, R., Rus, D. L., & Grosu, R. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9196608 · Cited by 28
“For instance, [20], [21] proposed to stabilize a trajectory controller realized by a Gaussian mixture regression, by introducing a stability condition based on a Lyapunov function.”
BackgroundMethodA Dynamic Behavior Control Framework for Physical Human-Robot Interaction
Liu, X., Ge, S. S., Zhao, F., & Mei, X. (2020) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-020-01286-x · Cited by 19
“It is assumed that the teaching trajectories satisfy the following autonomous ordinary differential equation (ODE) with first order [6, 18]: where ξ = [˙ x, x ] T , g : R d → R d denote the continuous differentiable function, which has a single equilibrium point ˙ ξ ∗ = g(ξ ∗ ) = 0, and (cid:6)…”
BackgroundLearning Stable Koopman Embeddings
Fan, F., Yi, B., Rye, D. C., Shi, G., & Manchester, I. R. (2022) · Proceedings of the ... American Control Conference/Proceedings of the American Control Conference · doi:10.23919/acc53348.2022.9867865 · Cited by 19
“Contraction is an important constraint for this data set as unconstrained models can have spurious attractors [5], leading to poor generalization to unseen initial conditions.”
BackgroundMethodHwang, C., Chen, B.-L., Syu, H.-T., Wang, C.-K., & Karkoub, M. A. (2014) · IEEE Systems Journal · doi:10.1109/jsyst.2014.2343236 · Cited by 20
“Khansari-Zadeh and Billard [5] decomposed the required motion into point-to-point actions for robots to imitate.”
BackgroundPeriodic DMP formulation for Quaternion Trajectories
Abu‐Dakka, F. J., Saveriano, M., & Peternel, L. (2021) · International Conference on Advanced Robotics (ICAR) · doi:10.1109/icar53236.2021.9659319 · Cited by 18
Jia, F., Jebelli, A., Ma, Y., & Ahmad, R. (2022) · Applied Sciences · doi:10.3390/app12115697 · Cited by 21
Zuo, G., Zhou, J., Gong, D., & Huang, G. (2023) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2023.3277102 · Cited by 18
“In this article, the desired position, desired velocity, actual position, and actual velocity are denoted by dre(t), vre(t), d(t), and v(t), respectively [28], [29].”
MethodA structured prediction approach for robot imitation learning
Duan, A., Batzianoulis, I., Camoriano, R., Rosasco, L., Pucci, D., & Billard, A. (2023) · The International Journal of Robotics Research · doi:10.1177/02783649231204656 · Cited by 27
“The algorithmic foundations of robot imitation learning notably include regression techniques (see Table 1) such as linear regression with basis functions (e.g., DMP and ProMP), kernel ridge regression (e.g., KMP), Gaussian mixture regression (e.g., SEDS), and Gaussian process regression (e.g., LGP and GPM), just to n…”
BackgroundSelf-citationLauretti, C., Tamantini, C., & Zollo, L. (2024) · IEEE Access · doi:10.1109/access.2023.3349093 · Cited by 20
“Other approaches based on the Gaussian Mixture Model (GMM) and the Gaussian Mixture Regression (GMR) resort to probabilistic methods for merging multiple demonstrations and building a representative trajectory [31], [32], [33].”
Generalizing demonstrated motion trajectories using coordinate-free shape descriptors
Vochten, M., Laet, T. D., & Schutter, J. D. (2019) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2019.103291 · Cited by 28
Diffeomorphically Learning Stable Koopman Operators
Bevanda, P., Beier, M., Kerz, S., Lederer, A., Sosnowski, S., & Hirche, S. (2022) · IEEE Control Systems Letters · doi:10.1109/lcsys.2022.3184927 · Cited by 40
“For the evaluation of the proposed Koopmanizing Flows,4 we compare its performance to the related work of SKEL5 [5] on the real-world LASA6 handwriting dataset [10], commonly used to compare approaches for learning stable dynamical systems.”
BackgroundMethodLearning modular policies for robotics
Neumann, G. M., Daniel, C., Paraschos, A., Kupcsik, A., & Peters, J. (2014) · Frontiers in Computational Neuroscience · doi:10.3389/fncom.2014.00062 · Cited by 29
“Movement primitives provide a compact representation of elemental movements by either parameterizing the trajectory (Schaal et al., 2003; Neumann, 2011; Rozo et al., 2013), muscle activation profiles (dAvella and Pai, 2010) or directly the control policy (Khansari-Zadeh and Billard, 2011).”
BackgroundMethodWu, Y. J., Su, Y., & Demiris, Y. (2014) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2014.05.010 · Cited by 22
“In particular, variants of GMR have been shown to be useful in many LbD applications, such as HMMGMR [1] and Dirichlet Process GMR (DP-GMR) [13].”
MethodTrajectory-Based Skill Learning Using Generalized Cylinders
Ahmadzadeh, S. R., & Chernova, S. (2018) · Frontiers in Robotics and AI · doi:10.3389/frobt.2018.00132 · Cited by 23
“Khansari-Zadeh and Billard (2011) introduced the Stable Estimator of Dynamical Systems (SEDS) approach which uses a constrained optimization technique to model a set of demonstrations as a dynamic system.”
BackgroundInfluentialA Probabilistic Representation for Dynamic Movement PrimitivesPreprint
Meier, F., & Stefan Schaal, A. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1612.05932 · Cited by 18
Nozaki, T., Mizoguchi, T., & Ohnishi, K. (2013) · IEEJ Journal of Industry Applications · doi:10.1541/ieejjia.3.1 · Cited by 17
Yuan, Y., Yu, Z. L., Hua, L., Cheng, Y., Li, J., & Sang, X. (2022) · Applied Intelligence · doi:10.1007/s10489-022-03219-7 · Cited by 17
“Moreover, in [23, 24], Khansari-Zadeh and Muelling proposed a mixture of movement primitives (MoMP) method to enable robots to select appropriate movement primitives as well as to generalize between them.”
MethodLearning Robust Point-to-Point Motions Adversarially: A Stochastic Differential Equation Approach
Zhang, H., Cheng, L., & Zhang, Y. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3251190 · Cited by 17
“%) compared with some state-of-the-art methods [7], [8], [11], [13].”
MethodResultInfluentialManschitz, S., Gienger, M., Kober, J., & Peters, J. (2018) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2018.2792531 · Cited by 17
Calinon, S., & Lee, D. (2017) · Springer eBooks · doi:10.1007/978-94-007-7194-9_68-1 · Cited by 18
Adaptive Control for Robotic Manipulators
(2017) · CRC Press eBooks · doi:10.1201/9781315166056 · Cited by 17
Dexterous Manipulation Through Imitation Learning: A Survey
An, S., Meng, Z., Tang, C., Zhou, Y., Liu, T., Ding, F., Zhang, S., Mu, Y., Song, R., Zhang, W., et al. (2025) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2025.3646183 · Cited by 71
“In parallel, Lyapunov-based approaches synthesized control policies that provably satisfy safety specifications, constructing Lyapunov functions as certificates of stability [332], [333], [334].”
Fast Approximate Multioutput Gaussian Processes
Joukov, V., & Kulić, D. (2022) · IEEE Intelligent Systems · doi:10.1109/mis.2022.3169036 · Cited by 16
“@Hið ffiffiffi 2 p asebsexÞ @xk 1⁄4 k!ð ffiffiffi 2 p asebseÞ ði kÞ! Hi kð ffiffiffi 2 p asebsexÞ: (33) FIGURE 1.”
Intent inference in shared-control teleoperation system in consideration of user behavior
Wang, L., Li, Q., Lam, J., Wang, Z., & Zhang, Z. (2021) · Complex & Intelligent Systems · doi:10.1007/s40747-021-00533-4 · Cited by 24
Oh, H., Sasaki, H., Michael, B., & Matsubara, T. (2022) · Neural Networks · doi:10.1016/j.neunet.2022.11.008 · Cited by 24
Fuzzy dynamical system for robot learning motion skills from human demonstration
Teng, T., Gatti, M., Poni, S., Caldwell, D. G., & Chen, F. (2023) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2023.104406 · Cited by 17
STAP: Sequencing Task-Agnostic Policies
Agia, C., Migimatsu, T., Wu, J., & Bohg, J. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10160220 · Cited by 62
“Dynamic Movement Primitives (DMPs) [11–13] are a form of LfD that learns the parameters of dynamical systems encoding movements [14–17].”
BackgroundLearning of parametric coupling terms for robot-environment interaction
Gams, A., Deniša, M., & Ude, A. (2015) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2015.7363559 · Cited by 16
“Generalization across start states was also shown by [14] and [15].”
BackgroundSparse Control for Dynamic Movement Primitives
Wensing, P. M., & Slotine, J.-J. E. (2017) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2017.08.1789 · Cited by 19
“Experiments in learning stable attractors from demonstration (Khansari-Zadeh and Billard, 2011) can be cast as convex problems through a contraction viewpoint (Ravichandar and Dani, 2015).”
BackgroundBayesian uncertainty modeling for programming by demonstration
Umlauft, J., Fanger, Y., & Hirche, S. (2017) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2017.7989759 · Cited by 16
“The proposed methods of Dynamic Movement Primitives (DMPs) [3] and Stable Estimator of Dynamical Systems (SEDS) [4] provide additional convergence guarantees for point-to-point and show promising results.”
MethodUncertainty-based Human Motion Tracking with Stable Gaussian Process State Space Models
Pöhler, L., Umlauft, J., & Hirche, S. (2019) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2019.01.002 · Cited by 16
Skill Acquisition via Automated Multi-Coordinate Cost Balancing
Ravichandar, H. C., Ahmadzadeh, S. R., Rana, M. A., & Chernova, S. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8793762 · Cited by 21
“This experiment evaluates MCCB on the publicly available LASA human handwriting library [5], that consists of handwriting motions collected from pen input using a Tablet PC.”
MethodPapageorgiou, D., Dimeas, F., Kastritsi, T., & Doulgeri, Z. (2019) · Robotica · doi:10.1017/s0263574719001437 · Cited by 20
“Two commonly used DSs for encoding kinematic behaviors are the autonomous DSs, for example, the ones utilizing Gaussian mixture models, 17 and the DMPs.”
MethodA Robot 3C Assembly Skill Learning Method by Intuitive Human Assembly Demonstration
Cao, Z., Hu, H., Yang, X., & Lou, Y. (2019) · WRC Symposium on Advanced Robotics and Automation (WRC SARA) · doi:10.1109/wrc-sara.2019.8931930 · Cited by 17
“Some policy learning methods, such as SEDS [13] can combine both of them to generate robot motion trajectory.”
MethodFourier movement primitives: an approach for learning rhythmic robot skills from demonstrations
Kulak, T., Silvério, J., & Calinon, S. (2020) · Robotics: Science and Systems · doi:10.15607/rss.2020.xvi.056 · Cited by 17
Duarte, N. F., Chatzilygeroudis, K. I., Santos-Victor, J., & Billard, A. (2020) · IEEE International Conference on Development and Learning (ICDL) · doi:10.1109/icdl-epirob48136.2020.9278084 · Cited by 16
“Each DS is encoded using Gaussian Mixture Models (GMM) which de(cid:2)nes a joint distribution function ) over the data as mixture of K Gaussian distributions [25], where (cid:25) k , (cid:22) k , and (cid:6) k are, respectively, the prior component, mean, and covariance matrix of the k th…”
Self-citationLearning stochastically stable Gaussian process state–space models
Umlauft, J., & Hirche, S. (2020) · IFAC Journal of Systems and Control · doi:10.1016/j.ifacsc.2020.100079 · Cited by 16
“As the convergence to the correct final position is important in goal-directed motions, enforcing stability to Gaussian mixture models (GMMs) is studied in [21] and [33] and more general techniques are developed in [22], [11], [43] and [36].”
BackgroundLearning from Demonstration facilitates Human-Robot Collaborative task execution
Koskinopoulou, M., Piperakis, S., & Trahanias, P. E. (2016) · ACM/IEEE International Conference on Human-Robot Interaction (HRI) · doi:10.1109/hri.2016.7451734 · Cited by 20
“Various learning techniques and motion models have been proposed, including polynomial and spline-based methods [3], nonlinear regression techniques, time-dependent dynamical systems [4], [5], and autonomous dynamical systems [6], [7].”
BackgroundOnline Model Learning Algorithms for Actor-Critic Control
Grondman, I. (2015) · Research Repository (Delft University of Technology) · doi:10.4233/uuid:415e14fd-0b1b-4e18-8974-5ad61f7fe280 · Cited by 23
“, Khansari-Zadeh and Billard (2011)).”
Modeling Long-horizon Tasks as Sequential Interaction LandscapesPreprint
Pirk, S., Hausman, K., Toshev, A., & Khansari, M. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.04843 · Cited by 33
“Primitives can be used as building blocks that can be composed, for example by a state machine [27], to enable more advanced robot behavior.”
BackgroundOnline Model Learning for Shape Control of Deformable Linear Objects
Yang, Y., Stork, J. A., & Stoyanov, T. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981080 · Cited by 22
“However, these algorithms employ function approximators such as Gaussian processes [19] and Gaussian mixture models [36].”
MethodKinesthetic Learning Based on Fast Marching Square Method for Manipulation
Prados, A., Mora, A., López, B., Muñoz, J., Garrido, S., & Barber, R. (2023) · Applied Sciences · doi:10.3390/app13042028 · Cited by 15
“…knowledge. in the first experiment, taking data with the robot in a kinesthetic way without considering , 130, 100 ] cm and the endpoint is P f = [ 25, 87, 89 ] cm with a position error of P error = It can be observed in Fig.27b how 710 the velocity matrix is modified, eliminating velocity lines…”
BackgroundInfluentialGe, D., Zhao, H., Wang, Y., Li, D., Li, X., & Ding, H. (2023) · Robotics and Computer-Integrated Manufacturing · doi:10.1016/j.rcim.2023.102669 · Cited by 15
Full-Body Postural Control of a Humanoid Robot with Both Imitation Learning and Skill Innovation
González-Fierro, M., Balaguer, C., Swann, N., & Nanayakkara, T. (2014) · International Journal of Humanoid Robotics · doi:10.1142/s0219843614500121 · Cited by 18
“Several studies have been conducted in the area of learning from demonstration (LfD) using Gaussian mixture models (GMM), that encode a set of trajectories, and Gaussian mixture regression (GMR), to obtain a generalized version of these trajectories to perform a robot movement.(21) The problem of skill transfer and wh…”
MethodRecurrent Equilibrium Networks: Unconstrained Learning of Stable and Robust Dynamical Models
Revay, M., Wang, R., & Manchester, I. R. (2021) · IEEE Conference on Decision and Control (CDC) · doi:10.1109/cdc45484.2021.9683054 · Cited by 22
“, for polynomial models [10], [11], Gaussian mixture models [12], and recurrent neural networks [13], [14], [15], however the”
BackgroundVirtual reality for improving safety and collaborative control of industrial robots
Hernoux, F., Nyiri, É., & Gibaru, O. (2015) · Virtual Reality International Conference · doi:10.1145/2806173.2806197 · Cited by 18
Wu, R., Zhang, H., & Zhao, J. (2020) · IEEE Access · doi:10.1109/access.2020.3043524 · Cited by 16
“2) Trajectory and variable impedance profile learning and generalization based on demonstration data To let the robot learn human skills in different states, we combine the Gaussian mixture regression (GMR) and stable estimator of dynamical systems (SEDS) algorithms [35]– [37].”
MethodFast motion planning for a laboratory 3D gantry crane in the presence of obstacles
Vu, M. N., Zips, P., Lobe, A., Beck, F., Kemmetmüller, W., & Kugi, A. (2020) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2020.12.2427 · Cited by 25
A Variable Impedance Skill Learning Algorithm Based on Kernelized Movement Primitives
Liu, A., Zhan, S., Jin, Z., & Zhang, W. (2023) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2023.3250746 · Cited by 14
“Over the past decade, numerous LfD algorithms have been proposed, such as dynamic movement primitives (DMP) [6], [7], [8], [9], dynamic systems (DS) [10], [11], Gaussian mixture model (GMM) [12], [13], hidden Markov model (HMM) [14], [15], probabilistic movement primitives (ProMP) [16], [17], [18],…”
BackgroundMethodLiu, X., Wang, Z., Li, J., Cangelosi, A., & Yang, C. (2023) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2023.3288240 · Cited by 18
“1) Obstacle Avoidance Algorithm: Khansari-Zadeh and Billard [13] proposed a framework for obstacle avoidance based on superquadric potential functions and introduced the system speed when defining the potential.”
BackgroundInfluentialFusion dynamical systems with machine learning in imitation learning: A comprehensive overview
Hu, Y., Abu‐Dakka, F. J., Chen, F., Luo, X., Li, Z., Knoll, A., & Ding, W. (2024) · Information Fusion · doi:10.1016/j.inffus.2024.102379 · Cited by 25
“While IL has demonstrated its capacity to generate motion [18] and generalize to specific tasks like obstacle avoidance [66] and stability [95] through supervised learning [14] or unsupervised learning [106], it faces challenges in generalizing when dealing with limited dataset in new environments.”
BackgroundMethodInfluentialLearning from Demonstration Facilitates Human-Robot Collaborative Task Execution
Koskinopoulou, M., Piperakis, S., & Trahanias, P. E. (2016) · ACM/IEEE International Conference on Human-Robot Interaction (HRI) · doi:10.5555/2906831.2906843 · Cited by 18
Imitation learning-based framework for learning 6-D linear compliant motions
Suomalainen, M., Abu-dakka, F. J., & Kyrki, V. (2021) · Autonomous Robots · doi:10.1007/s10514-021-09971-y · Cited by 20
“This makes SEDS state-based learning.”
BackgroundMethodInfluentialModel-Based Reinforcement Learning For Robot Control
Li, X., Shang, W., & Cong, S. (2020) · International Conference on Advanced Robotics and Mechatronics (ICARM) · doi:10.1109/icarm49381.2020.9195341 · Cited by 20
“But the most effective model-based methods generally use comparatively simple function approximators, for instance, linear models [12] [13], Gaussian processes(GPs) [6] [7] and mixtures of Gaussians [14].”
MethodRobust Passivity-Based Dynamical Systems for Compliant Motion Adaptation
Huang, H., Guo, Y. P., Yang, G., Chu, J., Chen, X., Li, Z., & Yang, C. (2022) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2022.3166204 · Cited by 14
Optimized vision-based robot motion planning from multiple demonstrations
Shen, T., Radmard, S., Chan, A., Croft, E. A., & Chesi, G. (2017) · Autonomous Robots · doi:10.1007/s10514-017-9667-4 · Cited by 14
“Their features and drawbacks have already been compared in literature (Cali-non et al. 2010; Khansari-Zadeh and Billard 2011; Steinmetz 2014).”
Learning to Serve: An Experimental Study for a New Learning From Demonstrations Framework
Koc, O., & Peters, J. (2019) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2019.2896466 · Cited by 15
“trol Lyapunov-functions [13], and various other probabilistic approaches, such as the probabilistic movement primitives [14] or Gaussian mixture models [15].”
MethodLearning Feature Maps of the Koopman Operator: A Subspace Viewpoint
Lian, Y., & Jones, C. N. (2019) · IEEE Conference on Decision and Control (CDC) · doi:10.1109/cdc40024.2019.9029189 · Cited by 14
“In [12], a Gaussian mixture model (GMM) is applied to ensure the global stability where the learned dynamical system admits a Lyapunov function.”
MethodAW-Opt: Learning Robotic Skills with Imitation and Reinforcement at ScalePreprint
Lu, Y., Hausman, K., Chebotar, Y., Yan, M., Jang, E., Herzog, A., Xiao, T. G., Irpan, A., Khansari, M., Kalashnikov, D., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2111.05424 · Cited by 44
The Developments and Challenges Toward Dexterous and Embodied Robotic Manipulation: A Survey
Li, G., Wang, R., Xu, P., Ye, Q., & Chen, J. (2025) · IEEE Robotics & Automation Magazine · doi:10.1109/mra.2025.3642671 · Cited by 44
A Policy Searched-Based Optimization Algorithm for Obstacle Avoidance in Robot Manipulators
Liu, A., Fu, J., Zhan, S., Jin, Z., & Zhang, W. (2024) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2023.3344831 · Cited by 13
“The dynamical system (DS) algorithm proposed by Khansari et al. [14] also solves the abovementioned problems.”
Method
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.