Abstract:A Probabilistic Movement Primitive (ProMP) defines a distribution over trajectories with an associated feedback policy. ProMPs are typically initialized from human demonstrations and achieve task generalization through probabilistic operations. However, there is currently no principled guidance in the literature to determine how many demonstrations a teacher should provide and what constitutes a "good" demonstration for promoting generalization. In this paper, we present an active learning approach to learning… Show more
“…Aleatoric uncertainties represent the variations in the demonstrations, i.e., different possible ways to achieve a task. This is for example the uncertainty that is captured by probabilistic movement primitives (ProMPs) when fitting a Gaussian or a Gaussian mixture model (GMM) to the demonstrations [27]. Aleatoric uncertainties can typically be employed within a minimal intervention control strategy, where perturbations are corrected only if they have an impact on the task, which results in adaptive tracking gains that take into account the variations of the task [28,29].…”
Section: B Active Imitation Learningmentioning
confidence: 99%
“…Although Gaussian processes are efficient for capturing epistemic uncertainties (model uncertainties), they do not capture aleatoric uncertainties (variations of the task). In [27], an active learning method is proposed for learning movement primitives based on Gaussian mixture models. The context to query (final endeffector position) is selected based on the distance between this context and the different Gaussians of the mixture.…”
Section: B Active Imitation Learningmentioning
confidence: 99%
“…Lik. ): this method, similar to [27], chooses the goal that is the furthest from our current task representation. Formally, this means that we compute the marginal distribution of our BGMM over the goal space, and choose the goal that has the minimum likelihood under this distribution.…”
This article proposes an approach for coupling internally-guided learning and social interaction in the context of a multi-task robot skill acquisition framework. More specifically, we focus on learning a parametrized distribution of robot movement primitives by combining active intrinsically-motivated learning and active imitation learning. We focus on the case where the learning modalities to use are not specified in advance by the experimenter, but are chosen actively by the robot through experiences. Such approach aims at combining experiential and observational learning as efficiently as possible, by relying on a skill acquisition mechanism in which the agent/robot can orchestrate different learning strategies in an iterative manner, and modulate the use of these modalities based on previous experiences. We demonstrate the effectiveness of our approach on a waste throwing task with a simulated 7-DoF Franka Emika robot, where at each iteration of the learning process the robot can actively choose between observational/imitation learning and experiential/intrinsically-motivated learning.
“…Aleatoric uncertainties represent the variations in the demonstrations, i.e., different possible ways to achieve a task. This is for example the uncertainty that is captured by probabilistic movement primitives (ProMPs) when fitting a Gaussian or a Gaussian mixture model (GMM) to the demonstrations [27]. Aleatoric uncertainties can typically be employed within a minimal intervention control strategy, where perturbations are corrected only if they have an impact on the task, which results in adaptive tracking gains that take into account the variations of the task [28,29].…”
Section: B Active Imitation Learningmentioning
confidence: 99%
“…Although Gaussian processes are efficient for capturing epistemic uncertainties (model uncertainties), they do not capture aleatoric uncertainties (variations of the task). In [27], an active learning method is proposed for learning movement primitives based on Gaussian mixture models. The context to query (final endeffector position) is selected based on the distance between this context and the different Gaussians of the mixture.…”
Section: B Active Imitation Learningmentioning
confidence: 99%
“…Lik. ): this method, similar to [27], chooses the goal that is the furthest from our current task representation. Formally, this means that we compute the marginal distribution of our BGMM over the goal space, and choose the goal that has the minimum likelihood under this distribution.…”
This article proposes an approach for coupling internally-guided learning and social interaction in the context of a multi-task robot skill acquisition framework. More specifically, we focus on learning a parametrized distribution of robot movement primitives by combining active intrinsically-motivated learning and active imitation learning. We focus on the case where the learning modalities to use are not specified in advance by the experimenter, but are chosen actively by the robot through experiences. Such approach aims at combining experiential and observational learning as efficiently as possible, by relying on a skill acquisition mechanism in which the agent/robot can orchestrate different learning strategies in an iterative manner, and modulate the use of these modalities based on previous experiences. We demonstrate the effectiveness of our approach on a waste throwing task with a simulated 7-DoF Franka Emika robot, where at each iteration of the learning process the robot can actively choose between observational/imitation learning and experiential/intrinsically-motivated learning.
“…Dynamic Movement Primitives (DMPs) [12,33,29] have been widely used to perform diverse, dynamic tasks such as table tennis [22], panckake flipping [16] or tether-ball [25]. They are able to model smooth, natural motions, and have in fact been used to inspire many policy learning schemes [8,5,4,40,11,7]. More recent work [2,24,30,6] has shown DMPs can be incorporated in a differentiable, end-to-end deep learning setting, which is an attribute that H-NDPs leverage.…”
We tackle the problem of generalization to unseen configurations for dynamic tasks in the real world while learning from high-dimensional image input. The family of nonlinear dynamical system-based methods have successfully demonstrated dynamic robot behaviors but have difficulty in generalizing to unseen configurations as well as learning from image inputs. Recent works approach this issue by using deep network policies and reparameterize actions to embed the structure of dynamical systems but still struggle in domains with diverse configurations of image goals, and hence, find it difficult to generalize. In this paper, we address this dichotomy by leveraging embedding the structure of dynamical systems in a hierarchical deep policy learning framework, called Hierarchical Neural Dynamical Policies (H-NDPs). Instead of fitting deep dynamical systems to diverse data directly, H-NDPs form a curriculum by learning local dynamical system-based policies on small regions in state-space and then distill them into a global dynamical system-based policy that operates only from high-dimensional images. H-NDPs additionally provide smooth trajectories, a strong safety benefit in the real world. We perform extensive experiments on dynamic tasks both in the real world (digit writing, scooping, and pouring) and simulation (catching, throwing, picking). We show that H-NDPs are easily integrated with both imitation as well as reinforcement learning setups and achieve state-of-the-art results. Video results are at https://shikharbahl.github.io/hierarchical-ndps/.
“…To realise the immersive teleoperation, the design of multimodal interfaces is the premise. Recently, a number of [72,73] HMM Model the correlation between movement and sensory profiles. [74,75] HSMM Encode the duration information of each HMM state and robust to perturbation.…”
Section: The Design Of Multimodal Interfacementioning
Manipulation skill learning and generalisation have gained increasing attention due to the wide applications of robot manipulators and the spurt of robot learning techniques. Especially, the learning from demonstration method has been exploited widely and successfully in the robotic community, and it is regarded as a promising direction to realise the manipulation skill learning and generalisation. In addition to the learning techniques, the immersive teleoperation enables the human to operate a remote robot with an intuitive interface and achieve the telepresence. Thus, it is a promising way to transfer manipulation skills from humans to robots by combining the learning methods and teleoperation, and adapting the learned skills to different tasks in new situations. This review, therefore, aims to provide an overview of immersive teleoperation for skill learning and generalisation to deal with complex manipulation tasks. To this end, the key technologies, for example, manipulation skill learning, multimodal interfacing for teleoperation and telerobotic control, are introduced. Then, an overview is given in terms of the most important applications of immersive teleoperation platform for robot skill learning. Finally, this survey discusses the remaining open challenges and promising research topics. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.