Learning parametric dynamic movement primitives from multiple demonstrations

Matsubara, Takamitsu; Hyon, Sang-Ho; Morimoto, Jun

doi:10.1016/j.neunet.2011.02.004

Cited by 86 publications

(43 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The high dimensional action space could make the model learning and the optimal action search intractable since a huge number of training data are required. Applying a concept of muscle synergies [20] or other dimensionality reduction scheme [21] for the action space can be considered. Moreover, we may use the information from two or more sensors in the feature extraction as treated in e.g.…”

Section: Discussionmentioning

confidence: 99%

Object manifold learning with action features for active tactile object recognition

Tanaka

Matsubara

Ichien

et al. 2014

2014 IEEE/RSJ International Conference on Intelligent Robots and Systems

Self Cite

View full text Add to dashboard Cite

Abstract-In this paper, we consider an object recognition problem based on tactile information using a robot hand. The robot performs an exploratory action to the object to obtain the tactile information, however, poorly designed actions may not be sufficiently informative. In contrast, if we could collect sample data by sequentially performing informative actions, i.e., active learning, the required time would be drastically reduced. To this end, we propose a novel approach for active tactile object recognition. Our approach combines both an active learning scheme and a nonlinear dimensionality reduction method. We first extracts the object manifold, each coordinate of which represents an object, from tactile sensor data and action features using Gaussian Process Latent Variable Models. At the same time, a probabilistic model of the observed data related to the action and the object are learned. Then, with the learned model, optimally-informative exploratory actions can be computed sequentially, and performed to efficiently collect the data for recognition. We show experimental results that verify the effectiveness of our proposed method with synthetic data and a real robot.

show abstract

Section: Discussionmentioning

confidence: 99%

Object manifold learning with action features for active tactile object recognition

Tanaka

Matsubara

Ichien

et al. 2014

2014 IEEE/RSJ International Conference on Intelligent Robots and Systems

Self Cite

View full text Add to dashboard Cite

show abstract

“…Our future work includes to devise new ways to find a relation between the cost function and each DoF, in order to minimize further the loss of information when reducing the dimensionality. Also prior to learning, we can use several demonstrations to find a better representation of the relevant directions in the parameter space [26]. Moreover, we will study when to reduce dimensionality (e.g.…”

Section: Discussionmentioning

confidence: 99%

Dimensionality reduction and motion coordination in learning trajectories with Dynamic Movement Primitives

Colomé

Torras

2014

2014 IEEE/RSJ International Conference on Intelligent Robots and Systems

View full text Add to dashboard Cite

Abstract-Dynamic Movement Primitives (DMP) are nowadays widely used as movement parametrization for learning trajectories, because of their linearity in the parameters, rescaling robustness and continuity. However, when learning a movement with a robot using DMP, many parameters may need to be tuned, requiring a prohibitive number of experiments/simulations to converge to a solution with a locally or globally optimal reward.We propose here strategies to palliate this dimensionality problem: the first is to explore only along the most significant directions in the parameter space, and the second is to add a reduced second set of Gaussians that would optimize the trajectory after fixing the Gaussians approximating the demonstrated movement. Both strategies result in less Gaussian computations and better performance on learning algorithms.To further speed up the learning and allow for a better biased exploration, we also propose to coordinate the motion of different joints, by computing a coordination matrix initialized with the demonstrated movement and then automatically updating it by eliminating the degrees of freedom least affecting task performance.Our three proposals have been experimentally tested and the obtained results show that similar (or even better) performance can be obtained at a significantly lower computational cost by reducing the dimensionality of the exploration space.

show abstract

“…Several denominations have been introduced in the literature to describe these models, such as task-parameterized [40,11] (the denomination that will be used here), parametric [49,26,29] or stylistic [7]. In these models, the encoding of skills usually serve several purposes, including classification, prediction, synthesis and online adaptation.…”

Section: Adaptive Models Of Movementsmentioning

confidence: 99%

“…A taxonomy of task-parameterized models is presented in [8], classifying existing methods in three broad cate-gories: 1) Approaches employing M models for the M demonstrations, performed in M different situations, see e.g. [16,23,29,25,45,12,21]; 2) Approaches employing P models for the P frames of reference that are possibly relevant for the task, see e.g. [32,13]; 3) Approaches employing a single model whose parameters are modulated by task parameters, see e.g.…”

Section: Adaptive Models Of Movementsmentioning

confidence: 99%

Robot Learning with Task-Parameterized Generative Models

Calinon

2017

Springer Proceedings in Advanced Robotics

View full text Add to dashboard Cite

Task-parameterized models provide a representation of movement/behavior that can adapt to a set of task parameters describing the current situation encountered by the robot, such as location of objects or landmarks in its workspace. This paper gives an overview of the taskparameterized Gaussian mixture model (TP-GMM) presented in previous publications, and introduces a number of extensions and ongoing challenges required to move the approach toward unconstrained environments. In particular, it discusses its generalization capability and the handling of movements with a high number of degrees of freedom. It then shows that the method is not restricted to movements in task space, but that it can also be exploited to handle constraints in joint space, including priority constraints.

show abstract

Learning parametric dynamic movement primitives from multiple demonstrations

Cited by 86 publications

References 16 publications

Object manifold learning with action features for active tactile object recognition

Object manifold learning with action features for active tactile object recognition

Dimensionality reduction and motion coordination in learning trajectories with Dynamic Movement Primitives

Robot Learning with Task-Parameterized Generative Models

Contact Info

Product

Resources

About