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 101–200 of 772
Bayesian Gaussian Mixture Model for Robotic Policy Imitation
Pignat, E., & Calinon, S. (2019) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2019.2932610 · Cited by 62
“In this experiment, we are in a similar situation to [3], where we control the velocity of an object.”
BackgroundMethodInfluentialDynamical System Based Robotic Motion Generation With Obstacle Avoidance
Stavridis, S., Papageorgiou, D., & Doulgeri, Z. (2017) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2017.2651172 · Cited by 49
“Dynamical systems may be further used to encode rich motion patterns learned by demonstrations [3], [4].”
BackgroundStability-Guaranteed Reinforcement Learning for Contact-Rich Manipulation
Khader, S. A., Yin, H., Falco, P., & Kragić, D. (2020) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2020.3028529 · Cited by 60
“ies are predicted from a policy parameterized by Gaussian Mixutre Regression (GMR) model. Interestingly, stability was guaranteed for the trajectory generation loop by means of the method proposed in [23], but the overall stability was unclear because of the lack of unified analysis of both loops. The ANN based method…”
MethodA Robot Motion Learning Method Using Broad Learning System Verified by Small-Scale Fish-Like Robot
Xu, S., Xu, T., Li, D., Yang, C., Huang, C., & Wu, X. (2023) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2023.3269773 · Cited by 51
“In [9], a stable estimator of DS (SEDS) method was used to model the complex robot motions combined with the Lyapunov theory.”
BackgroundMethodInfluentialDistance based dynamical system modulation for reactive avoidance of moving obstacles
Saveriano, M., & Lee, D. (2014) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2014.6907685 · Cited by 48
Online Trajectory Planning in Dynamic Environments for Surgical Task Automation
Osa, T., Sugita, N., & Mitsuishi, M. (2014) · Robotics: Science and Systems X · doi:10.15607/rss.2014.x.011 · Cited by 45
Adapting to contacts: Energy tanks and task energy for passivity-based dynamic movement primitives
Shahriari, E., Kramberger, A., Gams, A., Ude, A., & Haddadin, S. (2017) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2017.8239548 · Cited by 47
“For example, Gaussian Mixture Models (GMMs) can be used [14], [15].”
MethodModeling of Human Welders’ Operations in Virtual Reality Human–Robot Interaction
Wang, Q., Jiao, W., Yu, R., Johnson, M. T., & Zhang, Y. (2019) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2019.2921928 · Cited by 49
“Gaussian mixture models (GMMs) [42], [43], hidden Markov models (HMMs) [41], [44] and conditional random fields (CRFs) [45] have been used for modeling human motion.”
MethodLearning Force‐Relevant Skills from Human Demonstration
Gao, X. D., Ling, J., Xiao, X., & Li, M. (2019) · Complexity · doi:10.1155/2019/5262859 · Cited by 53
“In these tasks, the skills to be learned are usually encoded in the trajectory level in terms of position and velocity profile [11–13].”
Neural learning of vector fields for encoding stable dynamical systems
Lemme, A., Neumann, K., Reinhart, F., & Steil, J. J. (2014) · Neurocomputing · doi:10.1016/j.neucom.2014.02.012 · Cited by 45
“Recent studies have emphasized that stability of a predefined target plays an important role for the generalization ability in robotics tasks [17].”
BackgroundMethodInfluentialAction sequence reproduction based on automatic segmentation and Object-Action Complexes
Wächter, M., Schulz, S., Asfour, T., Aksoy, E. E., Wörgötter, F., & Dillmann, R. (2013) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2013.7029975 · Cited by 44
“…that enables the generalization of DMPs to new situations using the available training movements and the goal of the task [9] and another approach using nonlinear dynamical systems with gaussian mixture models, which can respond immediately to perturbations encountered during the motion [10].”
MethodLearning stable Gaussian process state space models
Umlauft, J., Lederer, A., & Hirche, S. (2017) · Proceedings of the ... American Control Conference/Proceedings of the American Control Conference · doi:10.23919/acc.2017.7963165 · Cited by 44
“Alternatively, [1] uses Gaussian Mixture Models for learning stable dynamics, while [11] develops a generic framework for stabilizing learned dynamical systems using control Lyapunov functions.”
BackgroundMethodMotion2Vec: Semi-Supervised Representation Learning from Surgical Videos
Tanwani, A. K., Sermanet, P., Yan, A., Anand, R., Phielipp, M., & Goldberg, K. Y. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9197324 · Cited by 56
“of collecting/labelling videos at pixel level to reflect such associations can be very high for training vision-based deep models in robotics. The trajectory-centric invariant formulations as in [15], [50], [51] may not readily generalize to the video demonstrations. First, transforming the images with respect to an ar…”
BackgroundLearning stable dynamical systems using contraction theory
Blocher, C., Saveriano, M., & Lee, D. (2017) · International Conference on Ubiquitous Robots and Ambient Intelligence · doi:10.1109/urai.2017.7992901 · Cited by 54
Learning movement primitives for force interaction tasks
Kober, J., Gienger, M., & Steil, J. J. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139639 · Cited by 43
“The extracted MP s are learned by SEDS [25].”
BackgroundLearning From Demonstration and Interactive Control of Variable-Impedance to Cut Soft Tissues
Wu, R., & Billard, A. (2021) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2021.3123356 · Cited by 41
“This kind of model, similar to stable estimator of dynamical systems (SEDS) and locally active—globally stable dynamical systems (LAGS-DS), is proposed in [32] and [33].”
BackgroundMethodInfluentialSelf-citationTransfer learning in robotics: An upcoming breakthrough? A review of promises and challenges
Jaquier, N., Welle, M. C., Gams, A., Yao, K., Fichera, B., Billard, A., Ude, A., Asfour, T., & Kragić, D. (2024) · The International Journal of Robotics Research · doi:10.1177/02783649241273565 · Cited by 59
“In this sense, for example, fast learning from a small set of demonstrations was applied with nonlinear autonomous dynamical system (DS), which have the ability to generalize motions to unseen contexts (Khansari-Zadeh and Billard 2011).”
Self-citationExtracting low-dimensional control variables for movement primitives
Rueckert, E., Mundo, J., Paraschos, A., Peters, J., & Neumann, G. M. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139390 · Cited by 41
“Movement primitives (MPs) are a compact parametric description of a movement [17], [8], [9], [4].”
BackgroundKinesthetic teaching of a therapist's behavior to a rehabilitation robot
Fong, J., & Tavakoli, M. (2018) · International Symposium on Medical Robotics · doi:10.1109/ismr.2018.8333285 · Cited by 41
“sophisticated Gaussian modeling methods such as the Stable Estimator of Dynamical Systems (SEDS), which provides a global-asymptotically stable task model [17].”
MethodCalinon, S., & Lee, D. (2018) · Springer eBooks · doi:10.1007/978-94-007-6046-2_68 · Cited by 47
Neurorobotics: From Vision to Action
van der Smagt, P., Arbib, M. A., & Metta, G. (2016) · Springer handbooks · doi:10.1007/978-3-319-32552-1_77 · Cited by 41
Robot Learning with Task-Parameterized Generative Models
Calinon, S. (2017) · Springer proceedings in advanced robotics · doi:10.1007/978-3-319-60916-4_7 · Cited by 49
“An option is to encode both static and dynamic features in the mixture model to retrieve continuous behaviors [51, 39, 22].”
BackgroundGaussian Mixture Model for 3-DoF orientations
Kim, S., Haschke, R., & Ritter, H. J. (2016) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2016.10.002 · Cited by 43
Data-driven Koopman operators for model-based shared control of human–machine systems
Broad, A., Abraham, I., Murphey, T. D., & Argall, B. (2020) · The International Journal of Robotics Research · doi:10.1177/0278364920921935 · Cited by 51
“For example, Neural Networks (Williams et al. (2017)), Gaussian Processes (Nguyen-Tuong et al. (2009)), and Gaussian Mixture Models (Khansari-Zadeh and Billard (2011)) have all shown great promise in this area.”
BackgroundFederated Ensemble Model-Based Reinforcement Learning in Edge Computing
Wang, J., Hu, J. W., Mills, J., Min, G., Xia, M., & Georgalas, N. (2023) · IEEE Transactions on Parallel and Distributed Systems · doi:10.1109/tpds.2023.3264480 · Cited by 50
“…significantly higher sample efficiency than model-free RL, and has been successfully applied to robotics [18], video games [13], etc., using a variety of dynamics models including Gaussian processes [24], linear models [25], [26], mixtures of Gaussians [27], and Deep Neural Networks (DNNs) [28]–[30].”
MethodDo You Need a Hand? – A Bimanual Robotic Dressing Assistance Scheme
Zhu, J., Gienger, M., Franzese, G., & Kober, J. (2024) · IEEE Transactions on Robotics · doi:10.1109/tro.2024.3366008 · Cited by 62
“Although there exist methods that combine dynamical systems with GMM to guarantee convergence [53], such methods do not work for TPGMM.”
MethodLearned graphical models for probabilistic planning provide a new class of movement primitives
Rueckert, E., Neumann, G. M., Toussaint, M., & Maass, W. (2013) · Frontiers in Computational Neuroscience · doi:10.3389/fncom.2012.00097 · Cited by 52
“One such methods is the Stable Estimator of Dynamical Systems (Khansari-Zadeh and Billard, 2011) approach.”
MethodField, M., Stirling, D. A., Pan, Z. S., & Naghdy, F. (2015) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2015.2414277 · Cited by 41
“Optimizing the stability of a non-linear dynamical system derived from mixture models was addressed in [7] for pointto-point motion.”
BackgroundMethodInfluentialModel Predictive Motion Control based on Generalized Dynamical Movement Primitives
Krug, R., & Dimitrov, D. (2014) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-014-0100-3 · Cited by 39
“However, as stated by the authors in [14], with increasing number of DoF the learning problem can become intractable.”
BackgroundGeometrical Invariance and Smoothness Maximization for Task-Space Movement Generation
Meirovitch, Y., Bennequin, D., & Flash, T. (2016) · IEEE Transactions on Robotics · doi:10.1109/tro.2016.2581208 · Cited by 43
“3) Control Lyapunov function-based dynamic movements (CLF-DM)[50].”
Khadivar, F., & Billard, A. (2023) · IEEE Transactions on Robotics · doi:10.1109/tro.2023.3280028 · Cited by 57
Self-citationJin, Z., Si, W., Liu, A., Zhang, W., Yu, L. J., & Yang, C. (2023) · IEEE Transactions on Robotics · doi:10.1109/tro.2023.3303011 · Cited by 44
“A representative approach is the SEDS presented in [32], where the special structure of the GMM/GMR is utilized.”
BackgroundMethodVu, M. N., Lobe, A., Beck, F., Weingartshofer, T., Hartl-Nesic, C., & Kugi, A. (2022) · Control Engineering Practice · doi:10.1016/j.conengprac.2022.105255 · Cited by 37
Coordinated multi-arm motion planning: Reaching for moving objects in the face of uncertainty
Salehian, S. S. M., Figueroa, N., & Billard, A. (2016) · Robotics: Science and Systems · doi:10.15607/rss.2016.xii.019 · Cited by 39
Self-citationLearning-Based Kinematic Control Using Position and Velocity Errors for Robot Trajectory Tracking
Xu, S., Ou, Y., Wang, Z., Duan, J., & Li, H. (2020) · IEEE Transactions on Systems Man and Cybernetics Systems · doi:10.1109/tsmc.2020.3013904 · Cited by 36
“…time indexes, the corresponding desired velocities and accelerations at all the time instants are clearly known (as recorded in the LASA library [22], which contains detailed data of multiple handwritten letters and has been widely used as reference data in many learning-based control studies).”
BackgroundMethodResultInfluentialLiao, Z., Jiang, G., Zhao, F., Wu, Y., Yue, Y., & Mei, X. (2022) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2022.3212707 · Cited by 36
“…transferring approaches have been introduced, such as the dynamic movement primitive (DMP) [9], Gaussian mixture model [10], probabilistic movement primitive [11], task-parameterized Gaussian mixture model [12], stable estimator of dynamical system [13], and kernelized movement primitive [14].”
MethodThe Use of Hebbian Cell Assemblies for Nonlinear Computation
Tetzlaff, C., Dasgupta, S., Kulvičius, T., & Wörgötter, F. (2015) · Scientific Reports · doi:10.1038/srep12866 · Cited by 44
Khansari, M., Klingbeil, E., & Khatib, O. (2016) · The International Journal of Robotics Research · doi:10.1177/0278364916648389 · Cited by 35
“) with a set of autonomous or non-autonomous differential equations (Billard et al., 2008; Ijspeert et al., 2002; Khansari-Zadeh and Billard, 2011).”
BackgroundMethodInfluentialContinual learning from demonstration of robotics skills
Auddy, S., Hollenstein, J., Saveriano, M., Rodríguez-Sánchez, A. J., & Piater, J. (2023) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2023.104427 · Cited by 51
Quadruped Robots: Bridging Mechanical Design, Control, and Applications
Li, Q., Cicirelli, F., Vinci, A., Guerrieri, A., Qi, W., & Fortino, G. (2025) · Robotics · doi:10.3390/robotics14050057 · Cited by 53
Simultaneously learning actions and goals from demonstration
Akgün, B., & Thomaz, A. L. (2015) · Autonomous Robots · doi:10.1007/s10514-015-9448-x · Cited by 40
“Dynamical systemapproaches such as Stable Estimator of Dynamical Systems (SEDS) (Khansari-Zadeh and Billard 2011) and DynamicMovement Primitives (DMP) (Pastor et al. 2009) as well as mixture 1 We only assume that a segmentation for objects is available. models, such as Gaussian Mixture Models…”
BackgroundConstraints extraction from asymmetrical bimanual tasks and their use in coordinated behavior
Ureche, L. P., & Billard, A. (2018) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2017.12.011 · Cited by 45
“We further determined a continuous coupling function γ{νM ,νS} by encoding the norm of the coupled variables using a SEDS model (stable estimator of dynamical system [41]).”
BackgroundMethodInfluentialSelf-citationLearning position and orientation dynamics from demonstrations via contraction analysis
Ravichandar, H. C., & Dani, A. P. (2018) · Autonomous Robots · doi:10.1007/s10514-018-9758-x · Cited by 44
“Now, the estimate of the state derivative, ẋ, is given by expected value of the posterior density P (ẋn (t) |xn (t)) as follows (Cohn et al. 1996; Khansari-Zadeh and Billard 2011)”
BackgroundMethodInfluentialAdaptive obstacle avoidance in path planning of collaborative robots for dynamic manufacturing
Hu, Y., Wang, Y., Hu, K., & Li, W. (2021) · Journal of Intelligent Manufacturing · doi:10.1007/s10845-021-01825-9 · Cited by 35
A neural network based framework for variable impedance skills learning from demonstrations
Zhang, Y., Cheng, L., Cao, R., Li, H., & Yang, C. (2022) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2022.104312 · Cited by 39
Stability of Gaussian process state space models
Beckers, T. L., & Hirche, S. (2016) · European Control Conference (ECC) · doi:10.1109/ecc.2016.7810630 · Cited by 34
“Khansari-Zadeh et al. show for GMMs a syntheses approach for learning stable trajectories of a nonlinear dynamical system with GMMs [14].”
InfluentialA Physically-Consistent Bayesian Non-Parametric Mixture Model for Dynamical System Learning.
Figueroa, N., & Billard, A. (2018) · Infoscience (Ecole Polytechnique Fédérale de Lausanne) · Cited by 93
Self-citationWang, N., Chen, C., & Yang, C. (2019) · Neurocomputing · doi:10.1016/j.neucom.2019.04.100 · Cited by 35
Open-source benchmarking for learned reaching motion generation in robotics
Lemme, A., Meirovitch, Y., Khansari-Zadeh, M. S. M., Flash, T., Billard, A., & Steil, J. J. (2015) · Paladyn Journal of Behavioral Robotics · doi:10.1515/pjbr-2015-0002 · Cited by 47
Self-citationEvaluation of a probabilistic approach to learn and reproduce gestures by imitation
Calinon, S., Sauser, E. L., Billard, A., & Caldwell, D. G. (2010) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/robot.2010.5509988 · Cited by 35
Self-citationRemote Robotic Laboratories for Learning from Demonstration
Osentoski, S., Pitzer, B., Crick, C., Jay, G., Dong, S., Grollman, D. H., Suay, H. B., & Jenkins, O. C. (2012) · International Journal of Social Robotics · doi:10.1007/s12369-012-0157-8 · Cited by 32
“The learning method, called Stable Estimator of Dynamical Systems (SEDS) [21], models motions as a non-linear autonomous dynamical system and defines sufficient conditions to ensure global asymptotic stability at the target.”
MethodAccelerated Sensorimotor Learning of Compliant Movement Primitives
Petrič, T., Gams, A., Colasanto, L., Ijspeert, A. J., & Ude, A. (2018) · IEEE Transactions on Robotics · doi:10.1109/tro.2018.2861921 · Cited by 38
Robot trajectory tracking control using learning from demonstration method
Xu, S., Ou, Y., Duan, J., Wu, X., Feng, W., & Liu, M. (2019) · Neurocomputing · doi:10.1016/j.neucom.2019.01.052 · Cited by 35
Learning Dynamical Systems with Side Information
Ahmadi, A. A., & Khadir, B. E. (2023) · SIAM Review · doi:10.1137/20m1388644 · Cited by 40
Force Adaptation in Contact Tasks with Dynamical Systems
Amanhoud, W., Khoramshahi, M., Bonnesoeur, M., & Billard, A. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9197509 · Cited by 32
“Such dynamics can be learned from human demonstrations and locally modulated to meet these constraints [18], [24].”
BackgroundSelf-citationLocally active globally stable dynamical systems: Theory, learning, and experiments
Figueroa, N., & Billard, A. (2022) · The International Journal of Robotics Research · doi:10.1177/02783649211030952 · Cited by 44
“To tackle this problem, Khansari-Zadeh and Billard (2011) proposed the stable estimator of dynamical systems (SEDS), which approximates a non-linear DS as a combination of linear DSs, where g k ( j ) is a state-dependent mixing function and each linear DS is of the following form: f k ( j ) = A k j…”
BackgroundMethodInfluentialSelf-citationA computational analysis of motor synergies by dynamic response decomposition
Alessandro, C., Carbajal, J. P., & d’Avella, A. (2014) · Frontiers in Computational Neuroscience · doi:10.3389/fncom.2013.00191 · Cited by 32
“Moreover, at the current stage DRD does not provide proved stable controller, a feature that can be enjoyed in DMP and SEDS.”
BackgroundInfluentialStable Motion Primitives via Imitation and Contrastive Learning
Pérez‐Dattari, R., & Kober, J. (2023) · IEEE Transactions on Robotics · doi:10.1109/tro.2023.3289597 · Cited by 45
On handing down our tools to robots: Single-phase kinesthetic teaching for dynamic in-contact tasks
Montebelli, A., Steinmetz, F., & Kyrki, V. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139987 · Cited by 44
“In the current literature, many approaches explore the use of statistical [6], [9], [10] or dynamic models [11], [12], [13], [14], [15].”
BackgroundRey, J., Kronander, K., Farshidian, F., Buchli, J., & Billard, A. (2017) · Autonomous Robots · doi:10.1007/s10514-017-9636-y · Cited by 38
“Todealwiththis issue, we use results from Khansari-Zadeh and Billard (2011) to derive a second exploration strategy with stability guarantee.”
BackgroundMethodInfluentialSelf-citationKhoramshahi, M., & Billard, A. (2020) · Autonomous Robots · doi:10.1007/s10514-020-09934-9 · Cited by 30
Self-citationLearning Accurate and Stable Point-to-Point Motions: A Dynamic System Approach
Zhang, Y., Cheng, L., Li, H., & Cao, R. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3140677 · Cited by 35
Point cloud based dynamical system modulation for reactive avoidance of convex and concave obstacles
Saveriano, M., & Lee, D. (2013) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2013.6697135 · Cited by 30
Learning by Demonstration and Robust Control of Dexterous In-Hand Robotic Manipulation Skills
Solak, G., & Jamone, L. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8967567 · Cited by 32
“The most prominent ones of these differences are that in SEDS dynamical system is time-invariant and multi-dimensional while in DMP there is a temporal phase variable and each dimension is modeled with an independent system.”
BackgroundInfluentialProbabilistic Prioritization of Movement Primitives
Paraschos, A., Lioutikov, R., Peters, J., & Neumann, G. M. (2017) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2017.2725440 · Cited by 29
“MPs [2]–[6], are a powerful representation for encoding complex movements, much more flexible than using attractors or point-to-point movements, and enable adaptation of the learned movements without retraining.”
BackgroundXu, J., Xiao, M., & Ding, Y. (2019) · Science China Technological Sciences · doi:10.1007/s11431-018-9488-1 · Cited by 29
Robot complex motion learning based on unsupervised trajectory segmentation and movement primitives
Song, C., Liu, G., Zhang, X., Zang, X., Xu, C., & Zhao, J. (2019) · ISA Transactions · doi:10.1016/j.isatra.2019.08.007 · Cited by 32
Bioinspired Embodiment for Intelligent Sensing and Dexterity in Fine Manipulation: A Survey
Liu, Y., Li, Z., Liu, H., Kan, Z., & Xu, B. (2020) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2020.2971643 · Cited by 32
An Optimal Variable Impedance Control With Consideration of the Stability
Jin, Z., Liu, A., Zhang, W., & Yu, L. J. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3141759 · Cited by 29
Learning to Play Minigolf: A Dynamical System-Based Approach
Khansari-Zadeh, S. M., Kronander, K., & Billard, A. (2012) · Advanced Robotics · doi:10.1080/01691864.2012.728692 · Cited by 28
“Different motion models and subsequently different learning techniques have been proposed, including but not limited to: polynomial and spline-based methods [8–10], nonlinear regression techniques [11, 12], time-dependent DS [13, 14], and autonomous DS [2, 3, 15].”
BackgroundMethodInfluentialSelf-citationActive contextual policy search
Fabisch, A., & Metzen, J. H. (2014) · Journal of Machine Learning Research · doi:10.5555/2627435.2697072 · Cited by 38
“In this paper, we address how active task selection can be performed if a novel task is chosen after each trial which involves that typically no close-to-optimal policy has been learned in the last task.”
BackgroundPrediction of Intention during Interaction with iCub with Probabilistic Movement Primitives
Dermy, O., Paraschos, A., Ewerton, M., Peters, J., Charpillet, F., & Ivaldi, S. (2017) · Frontiers in Robotics and AI · doi:10.3389/frobt.2017.00045 · Cited by 29
“In the past decade, several frameworks have been proposed to represent movements primitives, frequently called skills, the most notable being Gaussian Mixture Models (GMM) [16, 17], Dynamic Movement Primitives (DMP) [18], Probabilistic Dynamic Movement Primitive (PDMP [19]) and Probabilistic Movement Primitives (ProMP…”
BackgroundKadi, H. A., & Terzić, K. (2023) · Sensors · doi:10.3390/s23052389 · Cited by 27
“Gaussian distributions are used in PoMPs [114] and SEDs [115] for better generalisation behaviour.”
Method“iCub, clean the table!” A robot learning from demonstration approach using deep neural networks
Kim, J., Cauli, N., Vicente, P., Damas, B., Cavallo, F., & Santos-Victor, J. (2018) · IEEE International Conference on Autonomous Robot Systems and Competitions · doi:10.1109/icarsc.2018.8374152 · Cited by 28
“particular task or skill, and have been extensively used in the recent years [16], [17], [21].”
BackgroundFrom Human Physical Interaction To Online Motion Adaptation Using Parameterized Dynamical Systems
Khoramshahi, M., Laurens, A., Triquet, T., & Billard, A. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8594366 · Cited by 27
“, where to polish or pick/place), we employ state-dependent dynamical systems [28].”
MethodSelf-citationDynamic Trajectory Generation and a Robust Controller to Intercept a Moving Ball in a Game Setting
Prakash, R., Behera, L., Mohan, S., & Jagannathan, S. (2019) · IEEE Transactions on Control Systems Technology · doi:10.1109/tcst.2019.2913129 · Cited by 29
“Other models of movement primitives including multiple demonstrations include hidden Markov model [11], Gaussian mixture model [12], [13], and learning skill manifolds [14].”
BackgroundThe Linear Programming Approach to Reach-Avoid Problems for Markov Decision Processes
Kariotoglou, N., Kamgarpour, M., Summers, T. H., & Lygeros, J. (2017) · Journal of Artificial Intelligence Research · doi:10.1613/jair.5500 · Cited by 27
“For a general class of MDPs modeled by Gaussian mixture kernels [35] we propose using Gaussian radial basis functions (GRBFs) for approximating the reach-avoid value functions.”
MethodNonlinear System Identification With Prior Knowledge on the Region of Attraction
Khosravi, M., & Smith, R. S. (2020) · IEEE Control Systems Letters · doi:10.1109/lcsys.2020.3005163 · Cited by 27
HIRL: Hierarchical Inverse Reinforcement Learning for Long-Horizon Tasks with Delayed RewardsPreprint
Krishnan, S., Garg, A., Liaw, R., Miller, L., Pokorny, F. T., & Goldberg, K. Y. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1604.06508 · Cited by 27
Koptev, M., Figueroa, N., & Billard, A. (2024) · The International Journal of Robotics Research · doi:10.1177/02783649241246557 · Cited by 26
Self-citationShavit, Y., Figueroa, N., Salehian, S. S. M., & Billard, A. (2018) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2018.2833497 · Cited by 26
“To ensure that the learnt models reach the desired task-space target, Dynamical System (DS) formulations have been used to ensure stability and convergence [8], [9].”
MethodSelf-citationKhansari, M., Kronander, K., & Billard, A. (2014) · Robotics: Science and Systems · doi:10.15607/rss.2014.x.022 · Cited by 26
“In this paper, we focus on developing a suitable mathematical representation for f(s, ṡ) so that it allows combining motion planning and interaction control with guaranteed global asymptotic stability.”
MethodSelf-citationLearning peg-in-hole assembly using Cartesian DMPs with feedback mechanism
Liu, N., Zhou, X., Liu, Z., Wang, H., & Cui, L. (2020) · Assembly Automation · doi:10.1108/aa-04-2020-0053 · Cited by 28
A Comparison of Imitation Learning Algorithms for Bimanual Manipulation
Drolet, M., Stepputtis, S., Kailas, S., Jain, A., Peters, J., Schaal, S., & Amor, H. B. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3445630 · Cited by 44
“In particular, imitation learning has been a useful strategy for learning various robotic tasks [6], [17], [18], especially manipulation-based tasks such as pick-and-place or grasping [19], [20], [21].”
MethodGaussian processes for dynamic movement primitives with application in knowledge-based cooperation
Fanger, Y., Umlauft, J., & Hirche, S. (2016) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2016.7759576 · Cited by 25
“In [3] a learning approach is presented, which constrains the leaning to stable motions employing Gaussion”
MethodHuman Motion Prediction in Human-Robot Handovers based on Dynamic Movement Primitives
Widmann, D., & Karayiannidis, Y. (2018) · European Control Conference (ECC) · doi:10.23919/ecc.2018.8550170 · Cited by 25
“They use a nonlinear autonomous dynamical system to describe a movement and provide sufficient conditions to ensure global asymptotic stability at the endpoint of the movement trajectory [7].”
BackgroundComposite dynamic movement primitives based on neural networks for human–robot skill transfer
Si, W., Wang, N., & Yang, C. (2021) · Neural Computing and Applications · doi:10.1007/s00521-021-05747-8 · Cited by 25
Constraint-based movement representation grounded in geometric features
Bartels, G., Kresse, I., & Beetz, M. (2013) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2013.7030027 · Cited by 28
“Representations in robot learning [3], [4] and control engineering, on the other hand, specify sophisticated movement control but only in terms of coordinate frames [5] or lowlevel state and control variables [6].”
BackgroundHumanoids skill learning based on real-time human motion imitation using Kinect
Elbasiony, R. M., & Gomaa, W. E. (2018) · Intelligent Service Robotics · doi:10.1007/s11370-018-0247-z · Cited by 27
“The proposed methods in [5–7, 13, 16 ]”
MethodPark, S., Lee, W., Chung, W. K., & Kim, K. (2018) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2018.2876676 · Cited by 25
“In order to learn the control inputs which accomplish the trapping task from successful demonstrations, we introduced a regression method, in particular GMR [16].”
MethodInfluentialA Novel Robust Imitation Learning Framework for Complex Skills With Limited Demonstrations
Wang, W., Zeng, C., Zhan, H. L., & Yang, C. (2024) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2024.3403833 · Cited by 34
“The Stable Estimator of Dynamical Systems (SEDS) [31] estimates nonlinear DS as a mixture of linear DS and parametrizes via GMM-GMR, rigorously guaranteeing global asymptotic stability through Lyapunov analysis.”
Learning a Predictive Model of Human Gait for the Control of a Lower-limb Exoskeleton
Aertbeliën, E., & Schutter, J. D. (2014) · IEEE/RAS-EMBS International Conference on Biomedical Robotics and Biomechatronics (BioRob) · doi:10.1109/biorob.2014.6913830 · Cited by 30
Khoramshahi, M., Henriks, G., Naef, A. C., Salehian, S. S. M., Kim, J.-Y., & Billard, A. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9196593 · Cited by 24
“Dynamical systems (DS) are effective tools to embed motion dynamics that resemble human dynamics and can be trained from human demonstrations [23].”
BackgroundSelf-citationStability Analysis and Controller Synthesis Using Single-Hidden-Layer ReLU Neural Networks
Samanipour, P., & Poonawala, H. A. (2023) · IEEE Transactions on Automatic Control · doi:10.1109/tac.2023.3270219 · Cited by 24
“Two obvious applications are in system identification, where the right-hand side of the dynamics models are learned from data [2], [3], [4], [5], [6], [7], and analysis, where Lyapunov functions are “learned” [8], [9], [10], [11].”
MethodLearning Skills From Demonstrations: A Trend From Motion Primitives to Experience Abstraction
Tavassoli, M., Katyara, S., Pozzi, M., Deshpande, N., Caldwell, D. G., & Prattichizzo, D. (2023) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2023.3296166 · Cited by 32
“Although, based on their tasks realization, MP approaches are divided into two main classes in [8]: 1) state-based MP [9], [10] and 2) trajectory-based MP [11], [12], [13].”
MethodA Passivity-Based Approach for Variable Stiffness Control With Dynamical Systems
Michel, Y., Saveriano, M., & Lee, D. (2023) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2023.3324141 · Cited by 24
“This is in addition to their flexibility in modeling a myriad of robotic tasks [19], [20], [21], and their ability to incorporate a wide range of machine learning algorithms, such as Gaussian mixtures models [22] and Gaussian Processes [23].”
BackgroundMethodInfluentialReal-Time Pose Imitation by Mid-Size Humanoid Robot With Servo-Cradle-Head RGB-D Vision System
Hwang, C., & Liao, G.-H. (2018) · IEEE Transactions on Systems Man and Cybernetics Systems · doi:10.1109/tsmc.2017.2783947 · Cited by 26
Dynamic Motion Planning for Autonomous Assistive Surgical Robots
Sozzi, A., Bonfè, M., Farsoni, S., Rossi, G. D., & Muradore, R. (2019) · Electronics · doi:10.3390/electronics8090957 · Cited by 26
Gaussian process movement primitive
Jin, Z., Liu, A., Zhang, W., Yu, L. J., & Yang, C. (2023) · Automatica · doi:10.1016/j.automatica.2023.111120 · Cited by 25
Control-oriented meta-learning
Richards, S. M., Azizan, N., Slotine, J.-J. E., & Pavone, M. (2023) · The International Journal of Robotics Research · doi:10.1177/02783649231165085 · Cited by 56
“Khansari-Zadeh and Billard (2011) and Medina and Billard (2017) train controllers to imitate known invertible dynamical systems while constraining the closed-loop system to be stable.”
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.