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 401–500 of 772
Learning Stable Dynamical Systems for Visual Servoing
Paolillo, A., & Saveriano, M. (2022) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra46639.2022.9811944 · Cited by 8
Willibald, C., & Lee, D. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981288 · Cited by 7
“Movement Primitives [1], [2] are an established approach to react to low-level task perturbations and allow to generalize trajectories to varying start and end points, while preserving the nature of the movement.”
BackgroundA novel approach to dynamic movement imitation based on quadratic programming
Cardoso, C., Jamone, L., & Bernardino, A. J. M. (2015) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2015.7139285 · Cited by 10
“One of the most promising approaches draws from the theory of dynamical systems, giving raise to solutions such as Stable Estimator of Dynamical Systems (SEDS [10]), sequenced Linear Dynamical Systems [11], Implicit Dynamical Systems [12] and, most notably, Dynamic Movement Primitives (DMPs [13], [14]).”
BackgroundRealtime Avoidance of Fast Moving Objects: A Dynamical System-based Approach
Khansari-Zadeh, S. M., & Billard, A. (2012) · Cited by 7
Self-citationVochten, M., Laet, T. D., & Schutter, J. D. (2016) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2016.7487139 · Cited by 15
“By setting the weighting matrices equal to the inverse of the covariance matrix along the trajectory, invariant features with low variance get a higher weight and are more likely to be maintained in the reconstruction.”
BackgroundGeneration of movements with boundary conditions based on optimal control theory
Herzog, S., Wörgötter, F., & Kulvičius, T. (2017) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2017.04.006 · Cited by 7
Learning Stabilizable Dynamical Systems via Control Contraction Metrics
Singh, S., Sindhwani, V., Slotine, J.-J. E., & Pavone, M. (2020) · Springer proceedings in advanced robotics · doi:10.1007/978-3-030-44051-0_11 · Cited by 8
Motor Skills Learning and Generalization with Adapted Curvilinear Gaussian Mixture Model
Zhang, H., & Leng, Y. (2019) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-019-00999-y · Cited by 7
“To this end, we use the goal-directed reaching dataset proposed in paper [14].”
MethodTowards Coordinated Robot Motions: End-to-End Learning of Motion Policies on Transform Trees
Rana, M. A., Li, A., Fox, D., Chernova, S., Boots, B., & Ratliff, N. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636097 · Cited by 7
“previous approaches for learning stable time-invariant policies [10], [18]–[20], however, are limited to learning motions associated with a single task assigned to a given robot body part (e.”
BackgroundLearning Self-Correctable Policies and Value Functions from Demonstrations with Negative SamplingPreprint
Luo, Y., Xu, H., & Ma, T. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1907.05634 · Cited by 10
“We refer to [Kurutach et al., 2018, Clavera et al., 2018, Sun et al., 2018b, Chua et al., 2018, Sanchez-Gonzalez et al., 2018, Pascanu et al., 2017, Khansari-Zadeh and Billard, 2011, Luo et al., 2018] and the reference therein for recent work on model-based RL. Off-policy reinforcement learning…”
ResultGlobally Stable Neural Imitation Policies
Abyaneh, A., Guzmán, M. S., & Lin, H.-C. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10610791 · Cited by 11
“…target endpoint, x ∗ ∈ X , and have zero velocity at the target, i.e., The mapping between states and actions can be modeled with a time-invariant autonomous DS [5], denoted by: In Equation (2), the function f corresponds to an ordinary differential equation describing the true underlying DS.”
BackgroundMethodA survey on learning an autonomous dynamic system for human–robot skills transfer from demonstration
Fu, J., Huang, H., Jin, Z., Liu, A., Zhang, W.-A., Yu, L. J., Si, W., & Yang, C. (2025) · Robotics and Computer-Integrated Manufacturing · doi:10.1016/j.rcim.2025.103092 · Cited by 8
Flow Policy: Generalizable Visuomotor Policy Learning via Flow Matching
Fang, Y., Zhang, X., Cheng, H., Zang, X., Song, R., & Zhao, J. (2025) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2025.3586019 · Cited by 19
“In imitation learning, a traditional approach uses dynamical systems to model trajectories [20], [21].”
From Movement Primitives to Distance Fields to Dynamical Systems
Li, Y., & Calinon, S. (2025) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2025.3595073 · Cited by 14
“Examples include SEDS [8], CLF-DM [9], Imitation Flow [10], and Neural Contractive Dynamical Systems (NCDS) [11], which all ensure stability and adaptability by learning or imposing constraints on the system dynamics.”
Constrained Dynamic Movement Primitives for Collision Avoidance in Novel Environments
Shaw, S., Jha, D. K., Raghunathan, A. U., Corcodel, R., Romeres, D., Konidaris, G., & Nikovski, D. N. (2023) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros55552.2023.10341839 · Cited by 6
“There is a large body of work that constrains more general parameterizable dynamical systems via Lyapunov functions [23], [24], [25] and contraction map-”
BackgroundSun, S., & Figueroa, N. (2024) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros58592.2024.10801844 · Cited by 11
A Novel Robotic Skill Learning Approach for Assembly Task With Dynamical System and Broad Learning
Zhang, J., Jin, Z., Zhao, Z., & Yang, C. (2025) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2025.3539329 · Cited by 9
“To ensure that all motions follow the demonstration closely and converge to the same attractor point, a stable estimator of dynamical systems (SEDSs) was proposed in [20] to learn the DS.”
Reaching and grasping kitchenware objects
de Figueiredo, R. P., Shukla, A., Aragão, D., Moreno, P., Bernardino, A. J. M., Santos-Victor, J., & Billard, A. (2012) · IEEE/SICE International Symposium on System Integration (SII) · doi:10.1109/sii.2012.6426944 · Cited by 6
“The distributions in 1) and 3) above are learned using the stable estimator of dynamical system (SEDS) technique [14] which ensures that the learned dynamical system has a single globally and asymptotically stable attractor.”
BackgroundSelf-citationDynamic attack motion prediction for kendo agent
Tanaka, Y., & Kosuge, K. (2014) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2014.6942857 · Cited by 6
Koskinopoulou, M., & Trahanias, P. (2016) · IEEE-RAS International Conference on Humanoid Robots · doi:10.1109/humanoids.2016.7803331 · Cited by 7
“Time-dependent dynamical systems [10], [11], autonomous dynamical systems [12], [13], polynomial and spline-based methods [14], nonlinear regression techniques are indicative examples of methods that have been used in the literature to tackle this problem.”
MethodLearning Compliant Manipulation Tasks from Force Demonstrations
Duan, J., Ou, Y., Xu, S., Wang, Z., Peng, A., Wu, X., & Feng, W. (2018) · IEEE International Conference on Cyborg and Bionic Systems (CBS) · doi:10.1109/cbs.2018.8612173 · Cited by 6
“Learning kinematic aspects of tasks by demonstrations [2]–[4] turned out to be efficient, simple and powerful.”
BackgroundGuan, Y. H., Wang, N., & Yang, C. (2021) · Frontiers in Neuroscience · doi:10.3389/fnins.2021.694914 · Cited by 8
“Very recently, Zeng et al. (2020) propose to use hidden semi-Markov models (HSMMs) and Gaussian mixture regression (GMR) to offer the capability of capturing the correlation between position, speed, and stiffness.”
BackgroundClosed-Loop Variable Stiffness Control of Dynamical Systems
Chen, X., Michel, Y., & Lee, D. (2021) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids47582.2021.9555795 · Cited by 6
Learning to Control Known Feedback Linearizable Systems From Demonstrations
Sultangazin, A., Pannocchi, L., Fraile, L., & Tabuada, P. (2023) · IEEE Transactions on Automatic Control · doi:10.1109/tac.2023.3272392 · Cited by 6
“Inparticular,Khansari-Zadehand Billard[25]discusslearninganonlinearautonomousdynamical systemthatbestfitstheexpertdemonstrations,whilebeingglob-allyasymptoticallystable.”
Generating Stable and Collision-Free Policies through Lyapunov Function Learning
Coulombe, A., & Lin, H.-C. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10160494 · Cited by 6
“However, [10], [11] relies on a chosen Lyapunov candidate function to be valid for the demonstrations.”
BackgroundMethodInfluentialLearning Coupled Dynamical Systems from human demonstration for robotic eye-arm-hand coordination
Lukic, L., Santos-Victor, J., & Billard, A. (2012) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2012.6651574 · Cited by 8
Self-citationEfficient policy search with a parameterized skill memory
Reinhart, F., & Steil, J. J. (2014) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2014.6942740 · Cited by 9
“25, 33615 Bielefeld, Germany, Email: {freinhar, jsteil}@cor-lab.uni-bielefeld.de, www.cor-lab.de has been proposed in [8], where a linear projection of DMP parameters is used to interpolate between reaching motions.”
BackgroundTool centered learning from demonstration for robotic arms with visual feedback
Phung, A. S., Malzahn, J., Hoffmann, F., & Bertram, T. (2012) · IEEE International Conference on Robotics and Biomimetics (ROBIO) · doi:10.1109/robio.2012.6491119 · Cited by 7
“On the other hand, the authors in [4], [5] propose a learning scheme for robot motion that rests upon an autonomous timeindependent dynamical system through Gaussian mixture models (GMM).”
BackgroundModified dynamic movement primitives for joining movement sequences
Kulvičius, T., Ning, K., Tamošiūnaitė, M., & Wörgötter, F. (2011) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2011.5979716 · Cited by 12
A Passivity-Based Strategy for Manual Corrections in Human-Robot Coaching
Landi, C. T., Ferraguti, F., Fantuzzi, C., & Secchi, C. (2019) · Electronics · doi:10.3390/electronics8030320 · Cited by 6
A Novel Semi-Autonomous Teleoperation Method for the TianGong-2 Manipulator System
Li, C., Jiang, Z., Li, Z., Fan, C., & Liu, H. (2019) · IEEE Access · doi:10.1109/access.2019.2952762 · Cited by 6
“As shown in [33], the task trajectories can be formulated into a first-order autonomous ordinary differential equation according to the dynamical system theory.”
MethodActive Improvement of Control Policies with Bayesian Gaussian Mixture Model
Girgin, H., Pignat, E., Jaquier, N., & Calinon, S. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9341187 · Cited by 7
“tion. B. Product of Experts Robot movements learned with state-action abstractions result in probabilistic controllers with no guarantee of stability, unless explicitly constrained to be stable as in [6]. To overcome this problem, we fuse the probabilistic unstable controller with another probabilistic stable controll…”
MethodDynamical system learning using extreme learning machines with safety and stability guarantees
Salehi, I., Rotithor, G., Yao, G., & Dani, A. P. (2021) · International Journal of Adaptive Control and Signal Processing · doi:10.1002/acs.3237 · Cited by 6
“11-13 In Reference 14, a non-linear state space system is learned using a stable estimator of dynamical systems algorithm that preserves the property of a global convergence to an equilibrium point.”
Reactive, task-specific object manipulation by metric reinforcement learning
Hangl, S., Uğur, E., Szedmák, S., Piater, J., & Ude, A. (2015) · International Conference on Advanced Robotics (ICAR) · doi:10.1109/icar.2015.7251511 · Cited by 6
“While DMPs can be used to encode a single motor primitive, in [5] Gaussian mixture models are used to encode multiple demonstrations within a single, stable dynamic system.”
BackgroundEmbodied robots driven by self-organized environmental feedback
Kubandt, F., Nowak, M., Koglin, T., Gros, C., & Sándor, B. (2019) · Adaptive Behavior · doi:10.1177/1059712319855622 · Cited by 6
“, Gaussian mixture models [20], where the parameters of the dynamical system are either uniquely de-”
BackgroundIncremental Learning Robot Task Representation and Identification
Zhou, X., Wu, H., Rojas, J., Xu, Z., & Li, S. (2020) · Springer eBooks · doi:10.1007/978-981-15-6263-1_3 · Cited by 6
Imitation Learning:A Survey of Learning Methods
Hussein, A., Gaber, M. M., Elyan, E., & Jayne, C. (2017) · ACM Computing Surveys · doi:10.1145/3071073 · Cited by 62
A Survey on Deep Generative Models for Robot Learning From Multimodal Demonstrations
Urain, J., Mandlekar, A., Du, Y., Shafiullah, N. M. “., Xu, D., Fragkiadaki, K., Chalvatzaki, G., & Peters, J. (2025) · IEEE Transactions on Robotics · doi:10.1109/tro.2025.3631816 · Cited by 67
Stable Inverse Reinforcement Learning: Policies From Control Lyapunov Landscapes
Tesfazgi, S., Sprandl, L., Lederer, A., & Hirche, S. (2024) · IEEE Open Journal of Control Systems · doi:10.1109/ojcsys.2024.3447464 · Cited by 6
“While some recent approaches concurrently learn a task-oriented potential function, e.g., a Lyapunov-like function, which is used to stabilize the learned dynamical system [27], [28], [29], the stabilization is only performed virtually and the control law is not applicable to a physical plant.”
BackgroundMethodDemonstration to Adaptation: A User-Guided Framework for Sequential and Real-Time Planning
Cai, K., Laha, R., Gong, Y., Chen, L., Zhang, L., Figueredo, L., & Haddadin, S. (2024) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros58592.2024.10802661 · Cited by 5
“Taking a probabilistic approach such as Stable Estimator of Dynamical Systems (SEDS), [6], or Probabilistic Movement Primitives (ProMPs), [7], improves generalization by exploring the variance information from multiple demonstrations.”
Deep Learning of Robotic Manipulator Structures by Convolutional Neural Network
Aref, M. M., & Mattila, J. (2018) · International Conference on Intelligent Control and Information Processing · doi:10.1109/icicip.2018.8606719 · Cited by 5
“Learning for control purposes and based on human skills have also been comprehensively studied [3].”
BackgroundA Wheeled Inverted Pendulum Learning Stable and Accurate Control from Demonstrations
Jin, S., & Ou, Y. (2019) · Applied Sciences · doi:10.3390/app9245279 · Cited by 5
“A stable estimator of a dynamic system (SEDS) was proposed in [27] with Gaussian mixture models (GMMs) to incorporate both accuracy and stability factors.”
MethodA Sampling-based Motion Planning Framework for Complex Motor Actions
Sobti, S., Shome, R., Chaudhuri, S., & Kavraki, L. E. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636395 · Cited by 7
Kernelized gradient descent method for learning from demonstration
Hu, K., Zhang, J., & Wu, D. (2023) · Neurocomputing · doi:10.1016/j.neucom.2023.126781 · Cited by 5
Sun, S., Gao, H., Li, T., & Figueroa, N. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3401128 · Cited by 6
An Adaptive Framework for Manipulator Skill Reproduction in Dynamic Environments
Donald, R., Hertel, B., Misenti, S., Yan, G., & Azadeh, R. (2024) · International Conference on Ubiquitous Robots (UR) · doi:10.1109/ur61395.2024.10597445 · Cited by 5
“Our LfD representation, Elastic-Laplacian Trajectory Editing (ELTE), combines ideas from Laplacian Trajectory Editing (LTE) [6] and elastic maps [7].”
BackgroundInfluentialStulp, F., Colomé, A., & Carme Torras, A. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10610581 · Cited by 5
Bayesian Optimization for Sample-Efficient Policy Improvement in Robotic Manipulation
Röfer, A., Nematollahi, I., Welschehold, T., Burgard, W., & Valada, A. (2024) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros58592.2024.10802400 · Cited by 9
“For a more detailed understanding of the inference procedure, please refer to [8].”
InfluentialLiu, Z., & Fang, Y. (2025) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2024.3524207 · Cited by 7
“Stable estimator of dynamical system is a dynamic system (DS)-based method, which can learn from multiple human demonstrations, and ensures global asymptotic stability at the target [11].”
SAFE-GIL: SAFEty Guided Imitation Learning for Robotic Systems
Ciftci, Y. U., Chiu, D., Feng, Z., Sukhatme, G. S., & Bansal, S. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11128298 · Cited by 21
Combined kinesthetic and simulated interface for teaching robot motion models
Cha, E., Kronander, K., & Billard, A. (2015) · IEEE RO-MAN · doi:10.1109/roman.2015.7333655 · Cited by 5
“This motion representation has several advantages such as a smooth generalization of the demonstrated motion as well as time-invariance which imply an inherent robustness to temporal and spatial perturbations as discussed in detail in [10] and [11].”
BackgroundSelf-citationNon-linear regression algorithms for motor skill acquisition: a comparison
Munzer, T., Stulp, F., & Sigaud, O. (2014) · Cited by 5
Nonparametric Bayesian Learning for Collaborative Robot Multimodal Introspection
Zhou, X., Wu, H., Rojas, J., Xu, Z., & Li, S. (2020) · Springer eBooks · doi:10.1007/978-981-15-6263-1 · Cited by 6
Shape-Preserving and Reactive Adaptation of Robot End-Effector Trajectories
Vochten, M., Decré, W., Aertbeliën, E., & Schutter, J. D. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2020.3048674 · Cited by 9
“Dynamical system approaches such as Dynamic Movement Primitives [8] or Stable Estimator of Dynamical Systems [9] adapt trajectories by calculating appropriate forcing terms in dynamical attractor systems.”
MethodBayesian Identification of Hamiltonian Dynamics from Symplectic Data
Galioto, N., & Gorodetsky, A. (2020) · IEEE Conference on Decision and Control (CDC) · doi:10.1109/cdc42340.2020.9303852 · Cited by 13
“A wide variety of recent work has focused on the same problem where a physics-informed identification of dynamical systems is sought by prioritizing adherence to physical laws and/or phenomena such as stability [1], [2], [3], [4], conservation of energy [5], [6], [7], the principle of least action [8], [9], [10], [11]…”
BackgroundCharacterizing Continuous Manipulation Families for Dexterous Soft Robot Hands
Sun, J., King, J. P., & Pollard, N. S. (2021) · Frontiers in Robotics and AI · doi:10.3389/frobt.2021.645290 · Cited by 5
Wang, Q., Zhao, S., & Wang, X. (2021) · Applied Sciences · doi:10.3390/app11219830 · Cited by 5
Real-time Whole-body Obstacle Avoidance for 7-DOF Redundant Manipulators
Zheng, D., Wu, X., & Pang, J. (2021) · Chinese Control and Decision Conference (CCDC) · doi:10.1109/ccdc52312.2021.9602238 · Cited by 5
Consensus-based Normalizing-Flow Control: A Case Study in Learning Dual-Arm Coordination
Yin, H., Verginis, C. K., & Kragić, D. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981827 · Cited by 5
“The trajectories are generated from linear parameter-varying systems with parameter constraints similar to [8] and an external allocation variable for synchronization.”
BackgroundMethodLearning Stabilization Control from Observations by Learning Lyapunov-like Proxy Models
Ganai, M., Hirayama, C., Chang, Y.-C., & Gao, S. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10160928 · Cited by 15
Computationally efficient neural hybrid automaton framework for learning complex dynamics
Wang, T., Yang, Y., & Xiang, W. (2023) · Neurocomputing · doi:10.1016/j.neucom.2023.126879 · Cited by 5
Automatic Curriculum Determination for Deep Reinforcement Learning in Reconfigurable Robots
Karni, Z., Simhon, O., Zarrouk, D., & Berman, S. (2024) · IEEE Access · doi:10.1109/access.2024.3406768 · Cited by 5
Similarity-Aware Skill Reproduction based on Multi-Representational Learning from Demonstration
Hertel, B., & Ahmadzadeh, S. R. (2021) · International Conference on Advanced Robotics (ICAR) · doi:10.1109/icar53236.2021.9659470 · Cited by 5
“6 shows similarity maps obtained from the three LfD representations for a 2D shape from the LASA dataset [6].”
BackgroundMethodInfluentialAcquisition of Behavioral Dynamics for Vision Based Mobile Robot Navigation from Demonstrations
Narayanan, K. K., Posada, L. F., Frank Hoffmann, B., & Bertram, T. (2013) · IFAC Proceedings Volumes · doi:10.3182/20130410-3-cn-2034.00045 · Cited by 5
Imitation learning with non-parametric regression
Vaandrager, M., Babuška, R., Buşoniu, L., & Lopes, G. A. D. (2012) · IEEE International Conference on Automation, Quality and Testing, Robotics (AQTR) · doi:10.1109/aqtr.2012.6237681 · Cited by 5
“hidden Markov models and Gaussian mixture models [7], [8]; the behavior can also be represented using so-called motor primitives [9].”
BackgroundLearning Stable Nonlinear Dynamical Systems with External Inputs using Gaussian Mixture Models
Ravichandar, H. C., Salehi, I., Baillie, B. P., Bollas, G. M., & Dani, A. P. (2018) · Proceedings of the ... American Control Conference/Proceedings of the American Control Conference · doi:10.23919/acc.2018.8431461 · Cited by 5
“Now, the estimate of the state derivative, ẋ, is given by expected value of the posterior density P (ẋn (t) |xn (t)) as follows [6], [16]”
BackgroundMethodOne-shot path planning for multi-agent systems using fully convolutional neural networkPreprint
Kulvičius, T., Herzog, S., Lüddecke, T., Tamošiūnaitė, M., & Wörgötter, F. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.00568 · Cited by 5
“s to generate trajectories using conventional methods such as splines [6,30] or more advanced state-of-the-art methods such as dynamic movement primitives (DMPs, [13]), Gaussian mixture models (GMMs, [16]), probabilistic movement primitives (PMPs, [26]) or optimal control primitives (OCPs, [12]). Note that in some cas…”
MethodRoCUS: Robot Controller Understanding via SamplingPreprint
Zhou, Y., Booth, S., Figueroa, N., & Shah, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.13615 · Cited by 7
Wu, H., Zhai, X., Zheng, H., Liao, Z., Xu, Z., & Zhou, X. (2025) · Journal of Manufacturing Processes · doi:10.1016/j.jmapro.2025.06.095 · Cited by 4
Excavator trajectory planning via global probabilistic learning from expert demonstrations
Feng, C., Zhang, Q., Wang, J., Liu, X., Shen, Y., Jia, Q., & Zhao, J. (2025) · Automation in Construction · doi:10.1016/j.autcon.2025.106736 · Cited by 4
Teaching Periodic Stable Robot Motion Generation via Sketch
Zhi, W., Tang, H., Zhang, T., & Johnson‐Roberson, M. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3518837 · Cited by 4
“Unlike approaches which learn end-effector policies which converged to a fixed point [19], [27], [28], we seek to enforce that the generated reference trajectories of this dynamical system, starting at different initial positions, to approach a surface and converge onto a cycle on the surface.”
Imitation learning-based Direct Visual Servoing using the large projection formulation
Auddy, S., Paolillo, A., Piater, J., & Saveriano, M. (2025) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2025.104971 · Cited by 8
“Such approaches successfully generate complex kinematic motion from previous demonstrations, see [6, 7, 8, 9].”
Hitting flying objects with learning from demonstration
Chen, J., Shen, S., & Lau, H. Y. K. (2017) · International Conference on Advanced Robotics (ICAR) · doi:10.1109/icar.2017.8023496 · Cited by 4
“In this work, first order dynamical systems (DS) [13] human demonstrations with with DS, all possible paths to embedded into a single model which simplifies the motion planning problem of hitting since the learned DS is time independent, temporal perturbations can be compens to hit flying objects.”
BackgroundA Dynamical System Approach to Real-time Three-Dimensional Concave Obstacle Avoidance
Zheng, D., Wu, X., Liu, Y., & Pang, J. (2020) · IEEE/ASME International Conference on Advanced Intelligent Mechatronics · doi:10.1109/aim43001.2020.9158904 · Cited by 4
“The original DS is learned using the stable estimator of dynamical systems (SEDS) method proposed in [20] based on a set of demonstrations.”
MethodA Framework for Transferring Surface Finishing Skills to New Surface Geometries
Kim, Y., Sloth, C., & Kramberger, A. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981163 · Cited by 4
“To generalize the learned patterns better than the mentioned LfD approaches, a learning framework named Stable Estimator of Dynamical Systems (SEDS) which is capable of ensuring stable trajectories without dependence on time has been introduced in [9].”
Modular Neural Network Policies for Learning In-Flight Object Catching with a Robot Hand-Arm System
Hu, W., Acero, F., Triantafyllidis, E., Liu, Z., & Li, Z. (2023) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros55552.2023.10341463 · Cited by 10
“Encoding the arm and hand as Dynamical Systems (DS) [7] is one approach to catching tasks [5], [6], where kinematically feasible trajectories of the robot arm and hand are collected as demonstrations.”
BackgroundRobust Learning from Demonstration Based on GANs and Affine Transformation
An, K., Wu, Z., Shangguan, Q., Song, Y., & Xu, X. (2024) · Applied Sciences · doi:10.3390/app14072902 · Cited by 4
Learning an Autonomous Dynamic System to Encode Periodic Human Motion Skills
Jin, Z., Liu, A., Zhang, W., Yu, L. J., & Yang, C. (2024) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2024.3397356 · Cited by 4
“In [14] and [15], an ADS was learned by using Lyapunov function (energy function) constrained GMM/GMR and neural network algorithms, respectively.”
On particle dispersion statistics using unsupervised learning and Gaussian mixture models
Christakis, N., & Drikakis, D. (2024) · Physics of Fluids · doi:10.1063/5.0229111 · Cited by 4
Lee, Y., Lee, B., Kim, S., & Park, F. C. (2025) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2025.3575313 · Cited by 9
“M OVEMENT primitives are mathematical models that encode trajectories offline and enable rapid online generation, typically fitted to data from human demonstrations [1], [2], [3], [4], [5], [6].”
Multimedia event detection with ℓ2-regularized logistic Gaussian mixture regression
Liu, C., Dong, S., Lu, B., & Abdel-Mottaleb, M. S. (2015) · Neural Computing and Applications · doi:10.1007/s00521-014-1810-y · Cited by 4
“[10] proposed a learning method, called Stable Estimator of Dynamical Systems (SEDS), to learn the parameters of a dynamical system (i.”
Learning by demonstration for constrained tasks
Papageorgiou, D., & Doulgeri, Z. (2020) · IEEE RO-MAN · doi:10.1109/ro-man47096.2020.9223579 · Cited by 4
“Such DS are the Dynamic Movement Primitives (DMP) [2], [3] and Gaussian Mixture Model (GMM) [4] both of which utilize function approximation methods to learn the kinematic behavior.”
A control scheme for haptic inspection and partial modification of kinematic behaviors
Papageorgiou, D., & Doulgeri, Z. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9341594 · Cited by 4
“The most popular dynamical systems for LfD are the Dynamic Movement Primitives (DMP) [5], [6] and Gaussian Mixture Models (GMMs) [7].”
BackgroundImitation Learning Study for Robotic Peg-in-hole Assembly
He, C., Fang, Z., Wu, G., Wang, J., & Pan, C. (2021) · International Symposium on Robotics & Intelligent Manufacturing Technology (ISRIMT) · doi:10.1109/isrimt53730.2021.9597099 · Cited by 4
Ti, B., Gao, Y., Shi, M., Fu, L., & Zhao, J. (2021) · International Journal of Advanced Robotic Systems · doi:10.1177/17298814211065577 · Cited by 4
“Khansari-Zadeh and Billard 27 ensured the global asymptotic stability of the generalized movement.”
BackgroundLearning Contraction Policies From Offline Data
Rezazadeh, N., Kolarich, M., Kia, S. S., & Mehr, N. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3145100 · Cited by 12
“For instance, [26], [27] consider learning stable dynamics models.”
BackgroundAn open-architecture integration solution for the research and development of smart robots
Do, T. T., Dinh, V. P., Nguyen, Q. H., Chu, D. H., & Ministry of Science and Technology (2022) · Ministry of Science and Technology Vietnam · doi:10.31276/vjste.64(4).36-44 · Cited by 4
Towards Imitation Learning of Dynamic Manipulation Tasks: A Framework to Learn from Failures
Langsfeld, J. D., Kaipa, K. N., Gentili, R. J., Reggia, J. A., & Gupta, S. K. (2014) · University Libraries (University of Maryland) · Cited by 4
Robot teaching by teleoperation based on visual interaction and neural network learning
Xu, Y., Yang, C., Zhong, J., Ma, H., Zhao, L., & Wang, M. (2017) · International Conference on Modelling, Identification and Control · doi:10.1109/icmic.2017.8321615 · Cited by 4
“The dataset is {x, ẋ}1 ,TL t=0 , ε is a zero mean gaussian noise [22].”
BackgroundIncremental Policy Refinement by Recursive Regression and Kinesthetic Guidance
Nemec, B., Simonič, M., Petrič, T., & Ude, A. (2019) · International Conference on Advanced Robotics (ICAR) · doi:10.1109/icar46387.2019.8981606 · Cited by 4
“Many previously presented MP schemes fulfill these requirements, i.e. Dynamic Motion Primitives (DMP) [10], Gaussian Mixture Models with Gaussian Mixture regression (GMM-GPR) [11], Probabilistic Motion Primitives (PMP), [12], Hidden Markov Models [7], [2], Interaction Primitives (IP) [13], etc.”
BackgroundLearning Stable Nonparametric Dynamical Systems with Gaussian Process Regression
Hirche, W. X. A. L. S. (2020) · mediaTUM – the media and publications repository of the Technical University Munich (Technical University Munich) · Cited by 4
Null space redundancy learning for a flexible surgical robot
Bruno, D., Calinon, S., & Caldwell, D. G. (2014) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2014.6907199 · Cited by 4
“This results into a novel application of GMR, which has been used in the literature for motion generation of autonomouse systems [8], [9], for coupling dynamical systems of dynamic movement primitives [10], [11] or to extract trajectories from time-indexed demonstrations [12].”
MethodFast Marching-based globally stable motion learning
Gómez, J. V., Álvarez, D., Garrido, S., & Moreno, L. E. (2015) · Soft Computing · doi:10.1007/s00500-015-1981-1 · Cited by 7
“However, a brief comparison against the SEDS method (Khansari-Zadeh and Billard 2011) is included.”
BackgroundMethodResultInfluentialOptimal trajectory generation for generalization of discrete movements with boundary conditions
Herzog, S., Wörgötter, F., & Kulvičius, T. (2016) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2016.7759486 · Cited by 4
“Trajectory generation methods play an important role in robotics and many different methods exist ranging from conventional methods such as splines [1] to more dynamic approaches such as dynamic movement primitives (DMPs, [2]), Gaussian mixture models (GMMs, [3]), and a recent framework called probabilistic movement p…”
MethodLearning Motion Trajectories from Phase Space Analysis of the Demonstration
Gesel, P., Begum, M., & LaRoche, D. P. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8794381 · Cited by 4
“The stable dynamics naturally enable generalization in new contexts, such as on-line trajectory adaptation and obstacle avoidance [7], [8], [18].”
BackgroundFong, J., Ocampo, R., & Tavakoli, M. (2019) · Springer eBooks · doi:10.1007/978-3-030-34230-2_10 · Cited by 7
Enhancing Robot Learning Through Cognitive Reasoning Trajectory Optimization Under Unknown Dynamics
Dong, Q., Wu, T., Zeng, P., Zang, C., Wan, G., & Cui, S. (2025) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2025.3558648 · Cited by 4
“Most existing meth-odsemploytask-specificpriors,includingthosefordynamicsys-tems[11],[12],policyparametersandstructures[13],[14],[15], expectedreturns[16],[17],[18],andthelearningofdynamic systemmodels[19],[20],toendowrobotswithinnatefounda-tionalknowledge.”
Yao, M. H., Shao, Z., Wang, Y., Wang, X., & Wang, L. (2025) · Advanced Engineering Informatics · doi:10.1016/j.aei.2025.103945 · Cited by 4
Multiscale Sensor Fusion and Continuous Control with Neural CDEs
Singh, S., Ramirez, F. M., Varley, J., Zeng, A., & Sindhwani, V. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9982210 · Cited by 7
Learning Pose Dynamical System for Contact Tasks under Human Interaction
Yang, S., Gao, X. D., Feng, Z., & Xiao, X. (2023) · Actuators · doi:10.3390/act12040179 · Cited by 3
“Dynamic Movement Primitives (DMP) [12] is a typically and widely used approach of the DS. DMP is implicitly time-dependent and requires heuristics to reset the phase variable in the face of temporal perturbations [13].”
BackgroundMethodInfluential
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.