Encoding of periodic and their transient motions by a single dynamic movement primitive

Ernesti, Johannes; Righetti, Ludovic; Do, Martin; Asfour, Tamim; Schaal, Stefan

doi:10.1109/humanoids.2012.6651499

Cited by 36 publications

(28 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The dynamical systems approach facilitates sequencing of primitives into complex actions [15]. Also, it was demonstrated in [16] that individual DMPs can encode both transient and rhythmic components of a movement. However, they are not (yet) modular (but see [17,18]), i.e.…”

Section: Movement Primitivesmentioning

confidence: 99%

“…multivariate Gaussians, eqns. (16) and (17)) and factorize between weights and MPs, Q(W :,: , Y : ) = Q(W :,: )Q(Y : ). Posterior parameters are in the following denoted with a tilde, e.g.χ i,s is the posterior mean of the weight that connects signal i with MP s. With these choices, it is straightforward (but tedious) to evaluate the variational lower bound L (eqn.…”

Section: The Morphable Movement Primitive Modelmentioning

confidence: 99%

See 1 more Smart Citation

A novel approach for the generation of complex humanoid walking sequences based on a combination of optimal control and learning of movement primitives

Clever

Harant

Koch

et al. 2016

Robotics and Autonomous Systems

View full text Add to dashboard Cite

Section: Movement Primitivesmentioning

confidence: 99%

Section: The Morphable Movement Primitive Modelmentioning

confidence: 99%

A novel approach for the generation of complex humanoid walking sequences based on a combination of optimal control and learning of movement primitives

Clever

Harant

Koch

et al. 2016

Robotics and Autonomous Systems

View full text Add to dashboard Cite

“…In this work, wiping movements are encoded as periodic DMP using a slight extension of the DMP formulation as suggested in [11] which allows the representation of a periodic motion as well as its corresponding discrete transient movement. In general, a DMP consists of two parts:…”

Section: A Wiping Skillmentioning

confidence: 99%

Learn to wipe: A case study of structural bootstrapping from sensorimotor experience

Do¹,

Schill²,

Ernesti³

et al. 2014

2014 IEEE International Conference on Robotics and Automation (ICRA)

Self Cite

View full text Add to dashboard Cite

In this paper, we address the question of generative knowledge construction from sensorimotor experience, which is acquired by exploration. We show how actions and their effects on objects, together with perceptual representations of the objects, are used to build generative models which then can be used in internal simulation to predict the outcome of actions. Specifically, the paper presents an experiential cycle for learning association between object properties (softness and height) and action parameters for the wiping task and building generative models from sensorimotor experience resulting from wiping experiments. Object and action are linked to the observed effect to generate training data for learning a nonparametric continuous model using Support Vector Regression. In subsequent iterations, this model is grounded and used to make predictions on the expected effects for novel objects which can be used to constrain the parameter exploration. The cycle and skills have been implemented on the humanoid platform ARMAR-IIIb. Experiments with set of wiping objects differing in softness and height demonstrate efficient learning and adaptation behavior of action of wiping.

show abstract

“…Other approaches have the goal of a safe physical interaction with a human user [3], goal convergence in constrained environments [4] or global convergence towards closed curves [5]. A recent trend focusing on imitation learning incorporates the modulation of a dynamical system [6,7], thus overcoming the problem of getting stuck in a local minimum and being able to encode more complex movements. An elaborate approach for convex-shaped obstacle avoidance while maintaining the shape of a goal-driven motion is presented in [8].…”

Section: Introductionmentioning

confidence: 98%

Laplacian Trajectory Vector Fields for Robotic Movement Imitation and Adaption

Nierhoff

Hirche

Nakamura

2014

Advances on Theory and Practice of Robots and Manipulators

View full text Add to dashboard Cite

Vector field methods for trajectory following are generally computed offline before execution and thus only applicable to static trajectories. In contrast this paper introduces Laplacian trajectory vector fields (LTVF) as a computationally efficient method for creating convergent vector fields towards a discretized reference trajectory. In case of environmental changes both the vector field and the reference trajectory can be quickly recomputed. The conducted experiment uses a HRP-4 robot in order to display the applicability to daily life problems.

show abstract

Encoding of periodic and their transient motions by a single dynamic movement primitive

Cited by 36 publications

References 13 publications

A novel approach for the generation of complex humanoid walking sequences based on a combination of optimal control and learning of movement primitives

A novel approach for the generation of complex humanoid walking sequences based on a combination of optimal control and learning of movement primitives

Learn to wipe: A case study of structural bootstrapping from sensorimotor experience

Laplacian Trajectory Vector Fields for Robotic Movement Imitation and Adaption

Contact Info

Product

Resources

About