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
Imitation Learning: A Survey of Learning Methods
Hussein, A., Gaber, M. M., Elyan, E., & Jayne, C. (2018) · ACM Computing Surveys · doi:10.1145/3054912 · Cited by 1,060
“Statistical models can deal with sensor error or inaccurate demonstrations; however, incomplete demonstrations can result in suboptimal behavior [Khansari-Zadeh and Billard 2011].”
BackgroundNeural Network Dynamics for Model-Based Deep Reinforcement Learning with Model-Free Fine-Tuning
Nagabandi, A., Kahn, G., Fearing, R. S., & Levine, S. (2018) · 2018 IEEE International Conference on Robotics and Automation (ICRA) · doi:10.1109/icra.2018.8463189 · Cited by 848
Progress and prospects of the human–robot collaboration
Ajoudani, A., Zanchettin, A. M., Ivaldi, S., Albu-Schäffer, A., Kosuge, K., & Khatib, O. (2018) · Autonomous Robots · doi:10.1007/s10514-017-9677-2 · Cited by 832
Recent Advances in Robot Learning from Demonstration
Ravichandar, H., Polydoros, A. S., Chernova, S., & Billard, A. (2020) · Annual Review of Control, Robotics, and Autonomous Systems · doi:10.1146/annurev-control-100819-063206 · Cited by 752
“Existing literature in LfD has explored a wide variety of choices for what is considered to be the state, such as end-effector position (81), velocity (70), orientation (82), force (66, 76), joint angles (75), and torques (79).”
BackgroundMethodInfluentialSelf-citationModel-based Reinforcement Learning: A Survey
M. Moerland, T., Broekens, J., Plaat, A., & M. Jonker, C. (2023) · Foundations and Trends in Machine Learning · doi:10.1561/2200000086 · Cited by 524
(2013) · Lincoln Repository (University of Lincoln) · Cited by 503
Control-limited differential dynamic programming
Tassa, Y., Mansard, N., & Todorov, E. (2014) · 2014 IEEE International Conference on Robotics and Automation (ICRA) · doi:10.1109/icra.2014.6907001 · Cited by 440
“Ad-hoc task trajectories can be learned [9], which enlarge the convergence basin with a-priori knowledge and provide a consistent way to define complex task trajectories, but this is difficult to generalize to new situations.”
BackgroundA tutorial on task-parameterized movement learning and retrieval
Calinon, S. (2016) · Intelligent Service Robotics · doi:10.1007/s11370-015-0187-9 · Cited by 422
An algorithmic perspective on imitation learning
Osa, T., Pajarinen, J., Neumann, G., Bagnell, J. A., Abbeel, P., & Peters, J. (2018) · Foundations and Trends in Robotics · doi:10.1561/2300000053 · Cited by 405
“Behavior cloning methods such as [Pomerleau, 1988, Khansari-Zadeh and Billard, 2011] also employ deterministic policies.”
BackgroundMethodInfluentialAn Algorithmic Perspective on Imitation Learning
Osa, T., Pajarinen, J., Neumann, G., Bagnell, J. A., Abbeel, P., & Peters, J. (2018) · now publishers, Inc. eBooks · doi:10.1561/9781680834116 · Cited by 368
Robot Learning System Based on Adaptive Neural Control and Dynamic Movement Primitives
Yang, C., Chen, C., He, W., Cui, R., & Li, Z. (2019) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2018.2852711 · Cited by 350
“The SEDS presented in [3] is similar to our DMP-based model.”
MethodInfluentialDeepMPC: Learning Deep Latent Features for Model Predictive Control
Lenz, I., Knepper, R., & Saxena, A. (2015) · Robotics: Science and Systems XI · doi:10.15607/rss.2015.xi.012 · Cited by 332
“Thus, several recent works attempt to learn more general dynamics models such as Gaussian mixture models [7, 21] and Gaussian processes [22].”
BackgroundDynamic movement primitives in robotics: A tutorial survey
Saveriano, M., Abu-Dakka, F. J., Kramberger, A., & Peternel, L. (2023) · The International Journal of Robotics Research · doi:10.1177/02783649231201196 · Cited by 267
“Notable approaches to learn stable and autonomous systems exploit Lyapunov theory (Khansari-Zadeh and Billard 2011, 2014), contraction theory (Ravichandar and Dani 2015; Blocher et al. 2017), diffeomorphic transformations (Neumann and Steil 2015; Perrin and Schlehuber-Caissier 2016), and passivity…”
BackgroundA dynamical system approach to realtime obstacle avoidance
Khansari-Zadeh, S. M., & Billard, A. (2012) · Autonomous Robots · doi:10.1007/s10514-012-9287-y · Cited by 264
“The original DS is defined in the joint space and is learned based on a set of demonstrations in the robot joint space using the SEDS learning algorithm.”
InfluentialSelf-citationTrajectories and keyframes for kinesthetic teaching: a human-robot interaction perspective
Akgun, B., Cakmak, M., Yoo, J. W., & Thomaz, A. L. (2012) · Proceedings of the seventh annual ACM/IEEE international conference on Human-Robot Interaction · doi:10.1145/2157689.2157815 · Cited by 237
“Dynamical system approaches such as Stable Estimator of Dynamical Systems (SEDS) [13] and Dynamic Movement Primitives (DMP) [17] and mixture models (e.”
MethodMohammad Khansari-Zadeh, S., & Billard, A. (2014) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2014.03.001 · Cited by 225
“In our previous work, Stable Estimator of Dynamical systems (SEDS) [6], we developed for the first time a formal stability analysis and formulated explicit constraints to ensure global asymptotic stability of nonlinear autonomous DS that are modeled as a mixture of Gaussian functions.”
BackgroundMethodResultInfluentialSelf-citationKim, S., Shukla, A., & Billard, A. (2014) · IEEE Transactions on Robotics · doi:10.1109/tro.2014.2316022 · Cited by 220
Self-citationVariable Impedance Control and Learning—A Review
Abu-Dakka, F. J., & Saveriano, M. (2020) · Frontiers in Robotics and AI · doi:10.3389/frobt.2020.590681 · Cited by 214
“Saveriano and Lee (2014) follow a similar approach, but exploit the constraints derived by Khansari-Zadeh and Billard (2011) to guarantee the convergence of the trajectory retrieved via GMR. Li et al. (2014) omitted the damping term from the interaction model and used GMM to encode the pose of the…”
BackgroundMethodInfluentialUsing probabilistic movement primitives in robotics
Paraschos, A., Daniel, C., Peters, J., & Neumann, G. (2018) · Autonomous Robots · doi:10.1007/s10514-017-9648-7 · Cited by 205
“We categorize MPs into state-based (Khansari-Zadeh and Billard 2011; Calinon et al. 2010a) and trajectory-based representations (Schaal et al. 2005; Neumann et al. 2009; Rückert et al. 2012; Rozo et al. 2013).”
BackgroundMethodInfluentialImproving Multi-Step Prediction of Learned Time Series Models
Venkatraman, A., Hebert, M., & Bagnell, J. (2015) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v29i1.9590 · Cited by 198
“…Vector Regression (Müller, Smola, and Rätsch 1997), Gaussian process regression (Wang, Hertzmann, and Blei 2005; Ko et al. 2007), Nadaraya-Watson kernel regression (Basharat and Shah 2009), Gaussian mixture models (Khansari-Zadeh and Billard 2011), and Kernel PCA (Ralaivola and D’Alche-Buc 2004).”
Keyframe-based Learning from Demonstration: Method and Evaluation
Akgun, B., Cakmak, M., Jiang, K., & Thomaz, A. L. (2012) · International Journal of Social Robotics · doi:10.1007/s12369-012-0160-0 · Cited by 197
Reinforcement Learning on Variable Impedance Controller for High-Precision Robotic Assembly
Luo, J., Solowjow, E., Wen, C., Ojea, J. A., Agogino, A. M., Tamar, A., & Abbeel, P. (2019) · 2019 International Conference on Robotics and Automation (ICRA) · doi:10.1109/icra.2019.8793506 · Cited by 183
“There are also approaches tackling this problem by explicitly modeling contact dynamics [29, 30, 31, 32, 33] Inoue et al.”
MethodBillard, A. G., Calinon, S., & Dillmann, R. (2016) · Springer Handbooks · doi:10.1007/978-3-319-32552-1_74 · Cited by 184
Self-citationSurvey of imitation learning for robotic manipulation
Fang, B., Jia, S., Guo, D., Xu, M., Wen, S., & Sun, F. (2019) · International Journal of Intelligent Robotics and Applications · doi:10.1007/s41315-019-00103-5 · Cited by 188
“This method obtains a nonlinear autonomous dynamic system capable of generating action-state decision (Khansari-Zadeh and Billard 2011).”
MethodRobot Learning from Human Teachers
Chernova, S., & Thomaz, A. L. (2014) · Synthesis Lectures on Artificial Intelligence and Machine Learning · doi:10.1007/978-3-031-01570-0 · Cited by 173
A survey of robot manipulation in contact
Suomalainen, M., Karayiannidis, Y., & Kyrki, V. (2022) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2022.104224 · Cited by 165
“Another method for exploiting GMM in a learning task is Stable Estimator of Dynamical Systems (SEDS) [195].”
MethodInfluentialHuman-in-the-Loop Optimization of Shared Autonomy in Assistive Robotics
Gopinath, D., Jain, S., & Argall, B. D. (2017) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2016.2593928 · Cited by 168
“In SEDS, the target poses are modeled as attractors of a dynamical system.”
InfluentialMuscle synergies in neuroscience and robotics: from input-space to task-space perspectives
Alessandro, C., Delis, I., Nori, F., Panzeri, S., & Berret, B. (2013) · Frontiers in Computational Neuroscience · doi:10.3389/fncom.2013.00043 · Cited by 162
“In most of these works modules are defined as kinematic-based controllers that are combined sequentially to obtain complex joint trajectories (Khansari-Zadeh and Billard, 2011; Ijspeert et al., 2013).”
BackgroundA Survey on Policy Search Algorithms for Learning Robot Controllers in a Handful of Trials
Chatzilygeroudis, K., Vassiliades, V., Stulp, F., Calinon, S., & Mouret, J.-B. (2020) · IEEE Transactions on Robotics · doi:10.1109/tro.2019.2958211 · Cited by 160
“There exist basically two types of trajectory-based policies: 1) way-point based policies [42], and 2) dynamical system based [35], [41].”
BackgroundMethodInfluentialLearning Compliant Manipulation through Kinesthetic and Tactile Human-Robot Interaction
Kronander, K., & Billard, A. (2013) · IEEE Transactions on Haptics · doi:10.1109/toh.2013.54 · Cited by 186
“The DS is modeled by a Gaussian Mixture Model, trained with the Stable Estimator of Dynamical Systems (SEDS) [33] algorithm to guarantee that the motion converges to the glass regardless of starting position.”
MethodSelf-citationStatistical dynamical systems for skills acquisition in humanoids
Calinon, S., Li, Z., Alizadeh, T., Tsagarakis, N. G., & Caldwell, D. G. (2012) · 2012 12th IEEE-RAS International Conference on Humanoid Robots (Humanoids 2012) · doi:10.1109/humanoids.2012.6651539 · Cited by 149
“The second includes dynamical systems (DS) that can swiftly react to continuous perturbations [5]– [7].”
BackgroundInfluentialCurrent Researches and Future Development Trend of Intelligent Robot: A Review
Wang, T., Tao, Y., & Liu, H. (2018) · Machine Intelligence Research · doi:10.1007/s11633-018-1115-1 · Cited by 170
“Supervised learning mainly focuses on robot′s learning by imitation and from demonstration[120, 121].”
BackgroundSilva, N., Zhang, D., Kulvičius, T., Gail, A., Barreiros, C., Lindstaedt, S., Kraft, M., Bölte, S., Poustka, L., Nielsen‐Saines, K., et al. (2021) · Research in Developmental Disabilities · doi:10.1016/j.ridd.2021.103854 · Cited by 165
Robot Skill Learning: From Reinforcement Learning to Evolution Strategies
Stulp, F., & Sigaud, O. (2013) · Paladyn, Journal of Behavioral Robotics · doi:10.2478/pjbr-2013-0003 · Cited by 132
Algorithmic Framework for Model-based Deep Reinforcement Learning with Theoretical Guarantees
Luo, Y., Xu, H., Li, Y., Tian, Y., Darrell, T. J., & Ma, T. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1807.03858 · Cited by 118
“…time-varying linear models (Levine & Koltun, 2013; Lioutikov et al., 2014; Levine & Abbeel, 2014; Yip & Camarillo, 2014), mixture of Gaussians (Khansari-Zadeh & Billard, 2011), to neural networks (Hunt et al., 1992; Nagabandi et al., 2017; Kurutach et al., 2018; Tangkaratt et al., 2014;…”
BackgroundXu, S., Liu, J., Yang, C., Wu, X., & Xu, T. (2021) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2021.3121080 · Cited by 134
“Based on the learning-based method and the classical DS model [3], [27], the speed correction value can be designed as”
BackgroundMethodResultInfluentialRobot learning of industrial assembly task via human demonstrations
Kyrarini, M., Haseeb, M., Ristić–Durrant, D., & Gräser, A. (2018) · Autonomous Robots · doi:10.1007/s10514-018-9725-6 · Cited by 152
“Another approach to model the robot motion with dynamic systems is Stable Estimator of Dynamic Systems ( SEDS ) (Khansari-Zadeh and Billard 2011).”
From inverse optimal control to inverse reinforcement learning: A historical review
Azar, N. A., Shahmansoorian, A., & Davoudi, M. (2020) · Annual Reviews in Control · doi:10.1016/j.arcontrol.2020.06.001 · Cited by 164
Robot Learning from Human Teachers
Chernova, S., & Thomaz, A. L. (2014) · Synthesis lectures on artificial intelligence and machine learning · doi:10.2200/s00568ed1v01y201402aim028 · Cited by 119
Interactive Language: Talking to Robots in Real Time
Lynch, C., Wahid, A., Tompson, J., Ding, T., Betker, J., Baruch, R., Armstrong, T., & Florence, P. (2022) · arXiv · doi:10.48550/arxiv.2210.06407 · Cited by 113
An Approach for Imitation Learning on Riemannian Manifolds
Zeestraten, M. J. A., Havoutis, I., Silvério, J., Calinon, S., & Caldwell, D. G. (2017) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2017.2657001 · Cited by 109
A dynamical system approach to task-adaptation in physical human–robot interaction
Khoramshahi, M., & Billard, A. (2018) · Autonomous Robots · doi:10.1007/s10514-018-9764-z · Cited by 142
“Our previous works (Khansari-Zadeh and Billard 2011; Kronander and Billard 2016) have demonstrate that dynamical systems as motion generators have a great capacity to encode for a task, generate smooth trajectories, and also comply at the level of motion-generation. we complement this body of work…”
BackgroundSelf-citationBiologically Inspired Motion Modeling and Neural Control for Robot Learning From Demonstrations
Yang, C., Chen, C., Wang, N., Ju, Z., Fu, J. Y., & Wang, M. (2018) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2018.2866477 · Cited by 112
“The GMR has also been utilized to construct the DS model called the stable estimator of DSs for stable motions [20].”
MethodA review on manipulation skill acquisition through teleoperation‐based learning from demonstration
Si, W., Wang, N., & Yang, C. (2021) · Cognitive Computation and Systems · doi:10.1049/ccs2.12005 · Cited by 145
“This DS‐based approach was employed to model various motions, such as playing minigolf [49], human handwriting motion [48].”
BackgroundMethodOn improving the extrapolation capability of task-parameterized movement models
Calinon, S., Alizadeh, T., & Caldwell, D. G. (2013) · 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems · doi:10.1109/iros.2013.6696414 · Cited by 100
A Dynamical System Approach for Softly Catching a Flying Object: Theory and Experiment
Salehian, S. S. M., Khoramshahi, M., & Billard, A. (2016) · IEEE Transactions on Robotics · doi:10.1109/tro.2016.2536749 · Cited by 132
“[23] S. Kim and A. Billard, “Estimating the non-linear dynamics of free-flying objects,” Robot.”
BackgroundMethodInfluentialSelf-citationOnline Trajectory Planning and Force Control for Automation of Surgical Tasks
Osa, T., Sugita, N., & Mitsuishi, M. (2017) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2017.2676018 · Cited by 107
“Generalization of demonstrated trajectories is one of the primary problems in LfD. Khansari-Zadeh and Billard [12], [13] developed a scheme to model demonstrated motions as a time-invariant dynamic system using Gaussian mixture models (GMMs).”
BackgroundMethodInfluentialAvoidance of Convex and Concave Obstacles With Convergence Ensured Through Contraction
Huber, L., Billard, A., & Slotine, J.-J. E. (2019) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2019.2893676 · Cited by 103
“In contrast to classical path planning, the control law is closed form, hence requires no re-planning, and can ensure impenetrability of obstacles [1], [2].”
BackgroundSelf-citationCOBOT Applications—Recent Advances and Challenges
Taesi, C., Aggogeri, F., & Pellegrini, N. (2023) · Robotics · doi:10.3390/robotics12030079 · Cited by 100
Kou, J., Wang, Y., Chen, Z., Shi, Y., & Guo, Q. (2024) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2024.3453037 · Cited by 93
“Inthismode,theexoskeletonrequires onlysimplepositioncontrol,andtheexoskeletoniscontrolled totrackadesiredtrajectory,whichistypicallygeneratedby usingmotionmodelingandmachinelearning[6].”
Passive Interaction Control With Dynamical Systems
Kronander, K., & Billard, A. (2015) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2015.2509025 · Cited by 91
“We have exemplified it on DS learned using LMDS [6], but its applicability extends more generally to autonomous DS models such as SEDS [5] and ELM [7].”
BackgroundMethodInfluentialSelf-citationLearning Compliant Movement Primitives Through Demonstration and Statistical Generalization
Deniša, M., Gams, A., Ude, A., & Petrič, T. (2015) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2015.2510165 · Cited by 99
“Other representations of movement primitives that can be used with multiple demonstrations include hidden Markov models [19], Gaussian mixture models [20], [21], etc.”
BackgroundLearning compact parameterized skills with a single regression
Stulp, F., Raiola, G., Hoarau, A., Ivaldi, S., & Sigaud, O. (2013) · 2013 13th IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2013.7030008 · Cited by 85
“Methods for trajectory-level skill encoding, the focus of this paper, are able to leverage multiple demonstrated trajectories to determine relevant movement constraints [2], generalize across start states [3], [4], or for adaptation to varying task parameters [5]–[9].”
BackgroundMethodOnline learning of varying stiffness through physical human-robot interaction
Kronander, K., & Billard, A. (2012) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2012.6224877 · Cited by 86
“We plan to integrate the system in this paper with our previous work for modeling motions as autonomous dynamical systems [18] to allow robust teaching of tasks without time dependency.”
MethodSelf-citationLearning robot motions with stable dynamical systems under diffeomorphic transformations
Neumann, K., & Steil, J. J. (2015) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2015.04.006 · Cited by 86
“They became of special interest by focusing on the “what to imitate” problem [12, 13].”
BackgroundMethodInfluentialDynamic Movement Primitives: Volumetric Obstacle Avoidance Using Dynamic Potential Functions
Ginesi, M., Meli, D., Roberti, A., Sansonetto, N., & Fiorini, P. (2021) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-021-01344-y · Cited by 94
A survey on underactuated robotic systems: Bio-inspiration, trajectory planning and control
Liu, P., Huda, M. N., Sun, L. J., & Yu, H. (2020) · Mechatronics · doi:10.1016/j.mechatronics.2020.102443 · Cited by 96
Concept2Robot: Learning manipulation concepts from instructions and human demonstrations
Shao, L., Migimatsu, T., Zhang, Q., Yang, K., & Bohg, J. (2021) · The International Journal of Robotics Research · doi:10.1177/02783649211046285 · Cited by 80
“kind of dynamical system is analytically well understood and is widely utilized in the robotic community [21].”
BackgroundInfluentialCombining physics and deep learning to learn continuous-time dynamics models
Lutter, M., & Peters, J. (2023) · The International Journal of Robotics Research · doi:10.1177/02783649231169492 · Cited by 122
“2001), Gaussian Mixture Models (Calinon et al. 2010; Khansari-Zadeh and Billard 2011), Gaussian Processes (Kocijan et al.”
BackgroundMethodControl strategies for cleaning robots in domestic applications: A comprehensive review
Kim, J., Mishra, A. K., Limosani, R., Scafuro, M., Cauli, N., Santos-Victor, J., Mazzolai, B., & Cavallo, F. (2019) · International Journal of Advanced Robotic Systems · doi:10.1177/1729881419857432 · Cited by 92
The Classification and New Trends of Shared Control Strategies in Telerobotic Systems: A Survey
Li, G., Li, Q., Yang, C., Su, Y., Zuqiang, Y., & Wu, X. (2023) · IEEE Transactions on Haptics · doi:10.1109/toh.2023.3253856 · Cited by 86
“…(GMR) [118], theHidden(Semi-)MarkovModel(HMM/HSMM)[119],[120], theDynamicMovementPrimitives(DMP)[121],theStableEs-timatorofDynamicalSystems(SEDS)[122],theProbabilistic Movement Primitives (ProMP) [123], the Task-Parameterized GMM (TPGMM) [124], and the Kernelized Movement Prim-itives (KMP)…”
BackgroundDeep Lagrangian Networks for end-to-end learning of energy-based control for under-actuated systems
Lutter, M., Listmann, K. D., & Peters, J. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8968268 · Cited by 113
“For the function approximations standard machine learning techniques such as Linear Regression [7], [22], Gaussian Mixture Regression [23], [24], Gaussian Process Regression [25], [9], [26], Support Vector Regression [8], [27], feedforward- [28], [29], [30], [10] or recurrent neural networks [31] have been used.”
MethodKhansari-Zadeh, S. M., & Khatib, O. (2015) · Autonomous Robots · doi:10.1007/s10514-015-9528-y · Cited by 74
“Each movement primitive codes a behavior (such as reaching for a cup, swinging a golf club, etc.) with a set of autonomous or non-autonomous differential equations (Billard et al. 2008; Ijspeert et al. 2002; Khansari-Zadeh and Billard 2011).”
BackgroundInfluentialSelf-citationFast and Stable Learning of Dynamical Systems Based on Extreme Learning Machine
Duan, J., Ou, Y., Hu, J., Wang, Z., Jin, S., & Xu, C. (2017) · IEEE Transactions on Systems Man and Cybernetics Systems · doi:10.1109/tsmc.2017.2705279 · Cited by 73
“Although our experiments showed that a large library of robot motions can be modeled while satisfying these conditions, these global stability conditions might be too stringent to accurately model some complex motion” [12].”
BackgroundMethodInfluentialA Constrained DMPs Framework for Robot Skills Learning and Generalization From Human Demonstrations
Lu, Z., Wang, N., & Yang, C. (2021) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2021.3057022 · Cited by 81
Human–robot skill transmission for mobile robot via learning by demonstration
Li, J., Wang, J., Wang, S., & Yang, C. (2021) · Neural Computing and Applications · doi:10.1007/s00521-021-06449-x · Cited by 79
Recurrent Equilibrium Networks: Flexible Dynamic Models With Guaranteed Stability and Robustness
Revay, M., Wang, R., & Manchester, I. R. (2023) · IEEE Transactions on Automatic Control · doi:10.1109/tac.2023.3294101 · Cited by 158
“Fornonlinearmodels,therehas alsobeenasubstantialvolumeofresearchonstabilityguaran-tees,e.g.,forpolynomialmodels[25],[26],[27],[28],Gaussian mixturemodels [29], andRNNs [30], [31], [32], [33], [34].”
From dynamic movement primitives to associative skill memories
Pástor, P., Kalakrishnan, M., Meier, F., Stulp, F., Buchli, J., Theodorou, E. A., & Schaal, S. (2012) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2012.09.017 · Cited by 92
“This approach is in contrast to state–space based learning of nonlinear dynamic systems, as recently suggested by Khansari-Zadeh and Billard [39].”
ResultTowards Robust Skill Generalization: Unifying Learning from Demonstration and Motion Planning.
Rana, M. A., Mukadam, M., Ahmadzadeh, S. R., Chernova, S., & Boots, B. (2017) · Conference on Robot Learning · Cited by 104
Recurrent Neural Network based MPC for Process Industries
Lanzetti, N., Lian, Y. Z., Cortinovis, A., Domínguez, L. F., Mercangöz, M., & Jones, C. N. (2019) · European Control Conference (ECC) · doi:10.23919/ecc.2019.8795809 · Cited by 73
“For example, in [2] Gaussian mixture models (GMM) are used to model point-to-point motion by demonstrations.”
MethodTowards a generalized theory comprising digital, neuromorphic and unconventional computing
Jaeger, H. (2021) · Neuromorphic Computing and Engineering · doi:10.1088/2634-4386/abf151 · Cited by 92
Karami, M., & Wang, L. (2018) · Energy and Buildings · doi:10.1016/j.enbuild.2018.02.032 · Cited by 68
Murata, S., Namikawa, J., Arie, H., Sugano, S., & Tani, J. (2013) · IEEE Transactions on Autonomous Mental Development · doi:10.1109/tamd.2013.2258019 · Cited by 75
Learning from Demonstration (Programming by Demonstration)
Calinon, S. (2018) · Springer eBooks · doi:10.1007/978-3-642-41610-1_27-1 · Cited by 66
Learning stabilizable nonlinear dynamics with contraction-based regularization
Singh, S., Richards, S. M., Sindhwani, V., Slotine, J.-J. E., & Pavone, M. (2021) · The International Journal of Robotics Research · doi:10.1177/0278364920949931 · Cited by 64
“In this line of work, one assumes perfect knowledge and invertibility of the robot’s controlled dynamics to solve for the input that realizes this desirable closed-loop motion (Lemme et al., 2014; Khansari-Zadeh and Khatib, 2017; Ravichandar et al., 2017; Khansari-Zadeh and Billard, 2011; Medina and Billard, 2017).”
BackgroundHuman robot cooperation with compliance adaptation along the motion trajectory
Nemec, B., Likar, N., Gams, A., & Ude, A. (2017) · Autonomous Robots · doi:10.1007/s10514-017-9676-3 · Cited by 64
“Khansari-Zadeh and Billard (2011) have introduced a method of encoding motion as a nonlinear autonomous Dynamical System (DS) and sufficient conditions to ensure global asymptotic stability at the target.”
MethodKrishnan, S., Garg, A., Liaw, R., Thananjeyan, B., Miller, L., Pokorny, F. T., & Goldberg, K. (2019) · The International Journal of Robotics Research · doi:10.1177/0278364918784350 · Cited by 63
“This has the interpretation of fitting a locally linear regression to the data (refer to (Moldovan et al. 2015; Khansari-Zadeh and Billard 2011; Kruger et al. 2010; Krishnan* et al. 2015; Murali* et al. 2016) for details).”
MethodColome, A., & Torras, C. (2018) · IEEE Transactions on Robotics · doi:10.1109/tro.2018.2808924 · Cited by 60
“This fitting, which can use different regression models such as Gaussian mixture models [7], is then adapted to the environmental conditions in order to modify the robot’s behavior [8].”
MethodLearning Barrier Functions for Constrained Motion Planning with Dynamical Systems
Saveriano, M., & Lee, D. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8967981 · Cited by 61
“We consider 3 planar dynamical systems, namely a stable DS obtained with the approach in [2] (Fig.”
BackgroundMethodLearning Stable Deep Dynamics ModelsPreprint
Manek, G., & Kolter, J. Z. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2001.06116 · Cited by 125
“The methods also have applicability to non-linear systems, and several authors have used LMI analysis to learn nonlinear dynamical systems by constraining the linearization of the systems to have global Lyapunov functions [10, 2, 20], The point we want to emphasize from the above discussion, though, is that the task o…”
MethodAdaptive-Control-Oriented Meta-Learning for Nonlinear Systems
Richards, S. M., Azizan, N., Slotine, J.-J. E., & Pavone, M. (2021) · Robotics: Science and Systems · doi:10.15607/rss.2021.xvii.056 · Cited by 93
Learning in-contact control strategies from demonstration
Racca, M., Pajarinen, J., Montebelli, A., & Kyrki, V. (2016) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2016.7759127 · Cited by 67
Learning Via-Point Movement Primitives with Inter- and Extrapolation Capabilities
Zhou, Y., Gao, J., & Asfour, T. (2019) · 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8968586 · Cited by 61
“As described in [4] and [8], DS assumes that a robot motion is governed by an autonomous dynamical system defined over the robot state space.”
BackgroundInteraction learning for dynamic movement primitives used in cooperative robotic tasks
Kulvičius, T., Biehl, M., Aein, M. J., Tamošiūnaitė, M., & Wörgötter, F. (2013) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2013.07.009 · Cited by 60
“In addition to this, our approach can also be applied to other trajectory-shapingmethods (like GMMs [3–5]).”
MethodAvoiding Dense and Dynamic Obstacles in Enclosed Spaces: Application to Moving in Crowds
Huber, L., Slotine, J.-J. E., & Billard, A. (2022) · IEEE Transactions on Robotics · doi:10.1109/tro.2022.3164789 · Cited by 76
“In contrast to classical path planning, the control law is closed-form, hence requires no replanning, and can ensure impenetrability of obstacles [1], [2].”
BackgroundSelf-citationFast diffeomorphic matching to learn globally asymptotically stable nonlinear dynamical systems
Perrin, N., & Schlehuber-Caissier, P. (2016) · Systems & Control Letters · doi:10.1016/j.sysconle.2016.06.018 · Cited by 57
“We evaluated our approach on the LASA Handwriting Dataset, similarly to [3, 4, 6].”
ImitationFlow: Learning Deep Stable Stochastic Dynamic Systems by Normalizing Flows
Urain, J., Ginesi, M., Tateo, D., & Peters, J. (2020) · 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9341035 · Cited by 56
Dynamic Movement Primitives: Volumetric Obstacle Avoidance
Ginesi, M., Meli, D., Calanca, A., Dall’Alba, D., Sansonetto, N., & Fiorini, P. (2019) · International Conference on Advanced Robotics (ICAR) · doi:10.1109/icar46387.2019.8981552 · Cited by 56
“In last fifteen years, various LfD approaches (as Gaussian Mixture Models [5], Extreme Learning Machines [6], [7], and others [8]) have been developed in order to replicate human gestures.”
MethodRobot Policy Improvement With Natural Evolution Strategies for Stable Nonlinear Dynamical System
Hu, Y., Chen, G., Li, Z., & Knoll, A. (2022) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2022.3192049 · Cited by 57
“In particular, the time-invariant dynamical system in [7] and the constraints of stability condition according to the Lyapunov stability are taken into consideration in the context of learning.”
BackgroundMethodInfluentialNeural learning of stable dynamical systems based on data-driven Lyapunov candidates
Neumann, K., Lemme, A., & Steil, J. J. (2013) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2013.6696505 · Cited by 55
“This function is also a valid Lyapunov function for estimates produced by SEDS [13].”
BackgroundMethodInfluentialCompliant skills acquisition and multi-optima policy search with EM-based reinforcement learning
Calinon, S., Kormushev, P., & Caldwell, D. G. (2013) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2012.09.012 · Cited by 56
“Examples of models that can be reformulated in this way are the Gaussian mixture regression (GMR) [16, 17], the stable estimator of dynamical systems (SEDS) [18], the dynamic movement primitives (DMP) [19, 20, 21], the Takagi-Sugeno model (TSM) [22], or computational models combining irrotational and solenoidal vector…”
BackgroundEstimating the non-linear dynamics of free-flying objects
Kim, S., & Billard, A. (2012) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2012.05.022 · Cited by 55
Self-citationConcept2Robot: Learning Manipulation Concepts from Instructions and Human Demonstrations
Shao, L., Migimatsu, T., Zhang, Q., Yang, K., & Bohg, J. (2020) · Robotics: Science and Systems XVI · doi:10.15607/rss.2020.xvi.082 · Cited by 52
Steinmetz, F., Montebelli, A., & Kyrki, V. (2015) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2015.7363552 · Cited by 58
“tical models based on regression extract the main statistical features of the demonstration and use that information to reproduce the task [6], [7], [8].”
MethodRobotic Skill Learning for Precision Assembly With Microscopic Vision and Force Feedback
Qin, F., Xu, D., Zhang, D., & Li, Y. (2019) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2019.2909081 · Cited by 55
“The SEDS is used to learn the GMM from X train , which guarantees the global stability of dynamical system while maximizes the similarity between demonstration and replay [24].”
MethodFong, J., Ocampo, R., Gross, D. P., & Tavakoli, M. (2020) · Journal of Occupational Rehabilitation · doi:10.1007/s10926-020-09888-w · Cited by 52
Zhang, J., Zhao, H., Chen, K., Fei, G., Li, X., Wang, Y., Yang, Z., Zheng, S., Liu, S., & Ding, H. (2025) · Robotics and Computer-Integrated Manufacturing · doi:10.1016/j.rcim.2025.103021 · Cited by 67
Incremental motion learning with locally modulated dynamical systems
Kronander, K., Khansari, M., & Billard, A. (2015) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2015.03.010 · Cited by 64
Self-citationRobot teaching by teleoperation based on visual interaction and extreme learning machine
Xu, Y., Yang, C., Zhong, J., Wang, N., & Zhao, L. (2017) · Neurocomputing · doi:10.1016/j.neucom.2017.10.034 · Cited by 50
A Dynamical System Approach to Motion and Force Generation in Contact Tasks
Amanhoud, W., Khoramshahi, M., & Billard, A. (2019) · Robotics: Science and Systems · doi:10.15607/rss.2019.xv.021 · Cited by 50
“Such dynamics can be learned from human demonstrations and locally modulated to meet these constraints [21, 23].”
BackgroundSelf-citation
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.