Movement imitation with nonlinear dynamical systems in humanoid robots

By Auke Jan Ijspeert, Jun Nakanishi, Stefan Schaal. 2002

1,004 citations recorded by the source corpus.

Subfields: Biomedical Engineering, Cognitive Neuroscience, Social Psychology, Reinforcement Learning, Robotics & Control

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

In Cognitive Neuroscience

In Social Psychology

In Reinforcement Learning

In Robotics & Control

Citations in the corpus

Time-Contrastive Networks: Self-Supervised Learning from Video (2017, methodology): “used for learning robotic skills from expert demonstrations [4, 5, 6, 7] and can be split into two areas: be-”
Learning to Generalize Across Long-Horizon Tasks from Human Demonstrations (2020, background): “Imitation learning has been applied to multiple domains such as playing table tennis [22], Go [37] and video games [33], and driving autonomously [4, 31] In robotics, the idea of robots learning from human demonstrations have been extensively explored [36, 17, 23, 11, 14, 3, 5].”
Learning Complex Dexterous Manipulation with Deep Reinforcement Learning and Demonstrations (2017, methodology): “d) Combining RL with demonstrations: Methods based on dynamic movement primitives (DMPs) [36], [50], [21], [18] have been used to effectively combine demonstrations and RL to enable faster learning.”
Generalization Through Hand-Eye Coordination: An Action Space for Learning Spatially-Invariant Visuomotor Control (2021, background): “Imitation learning (IL) has been widely studied in the context of robotic manipulation tasks [2, 4, 5, 11, 31].”
An Algorithmic Perspective on Imitation Learning (2018, background): “In current imitation learning, there are performance guarantees, e.g., stability of DMPs [Ijspeert et al., 2002a, Schaal et al., 2004] and a proof of low error in DAGGER [Ross et al., 2011].”
Collective robot reinforcement learning with distributed asynchronous guided policy search (2016, background): “For example, the well-known dynamic movement primitive representation [8] has been widely used to generalize learned skills by adapting the goal state, but it inherently restricts the learning process to trajectory-centric behaviors.”
Path integral guided policy search (2016, methodology): “PI(2) has been successfully used to learn parameters of trajectory-centric policies such as dynamic movement primitives [19] in grasping [20], pick-and-place [21] and variable impedance control tasks [22].”
Path Integral Policy Improvement with Covariance Matrix Adaptation (2012, methodology): “So far, PI 2 has mainly been applied to policies represented as Dynamic Movement Primitives (DMPs) (Ijspeert et al., 2002), where θ determines the shape of the movement.”
BM: An iterative algorithm to learn stable non-linear dynamical systems with Gaussian mixture models (2010, methodology): “To address the instability issue, the Dynamic Motor Primitive (DMP) method [6] was proposed.”
Reinforcement learning of motor skills with policy gradients (2008, methodology): “, (1) spline-based trajectory plans and (2) nonlinear dynamic motor primitives introduced in Ijspeert et al. (2002). Spline-based trajectory planning is well known in the robotics literature, see e.g., Miyamoto et al. (1996) and Sciavicco and Siciliano (2007). A desired trajectory is represented as ”
Towards Deep Developmental Learning (2016, background): “example of such primitives are dynamic movement primitives (DMPs) [129], [130].”
Learning to select and generalize striking movements in robot table tennis (2012, methodology): “…and end positions such as reaching, pointing, grasping and striking movements) the canonical system is defined as τ ż = −αzz, (5) where τ is a temporal scaling factor and αz is a predefined constant which is chosen such that the behavior is stable (Ijspeert et al., 2002; Schaal et al., 2007).”
Towards learning hierarchical skills for multi-phase manipulation tasks (2015, background): “These motor primitive representations are easily adapted to different situations and can be learned in a straightforward manner [18], [33] .”
Learning motion primitive goals for robust manipulation (2011, methodology): “a DMP is trained with an observed trajectory through supervised learning [5].”
Task adaptation through exploration and action sequencing (2009, methodology): “[4] in the form of a linear combination of”
Hierarchical visuomotor control of humanoids (2018, background): “The study of movement primitives for robotics (c.f. “motor” primitives in neuroscience) descends from the work of Ijspeert et al. (2002).”
Generalized Cylinders for Learning, Reproduction, Generalization, and Refinement of Robot Skills (2017, background): “Finally, to maintain the shape of the movement during generalization, DMPs require significant tuning of continuous parameters, including those of the dynamical systems, such as time constants and scaling factors [15].”