Learning multiple strategies to perform a valve turning with underwater currents using an I-AUV

Carrera, Arnau; Palomeras, Narcís; Hurtós, Natàlia; Kormushev, Petar; Carreras, Marc

doi:10.1109/oceans-genova.2015.7271609

Cited by 18 publications

(20 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Kumagai et al (2016) devised a method for improving the localization system of a humanoid by fusing internal sensors and visual information to achieve better odometry. Carrera, Palomeras, Hurtós, Kormushev, and Carreras (2015) developed a method enabling an underwater robot to learn valve-turning strategies for different levels of water current from pilot demonstrations. Newman et al (2014) devised a method for fast, robust valve-turning using compliant control with force feedback.…”

Section: Related Workmentioning

confidence: 99%

Multimodal sensing and active continuous closed‐loop feedback for achieving reliable manipulation in the outdoor physical world

Chan¹,

Mizohana²,

Chen³

et al. 2018

Journal of Field Robotics

View full text Add to dashboard Cite

The use of field robots can greatly decrease the amount of time, effort, and associated risk compared to if human workers were to carryout certain tasks such as disaster response. However, transportability and reliability remain two main issues for most current robot systems. To address the issue of transportability, we have developed a lightweight modularizable platform named AeroArm. To address the issue of reliability, we utilize a multimodal sensing approach, combining the use of multiple sensors and sensor types, and the use of different detection algorithms, as well as active continuous closed‐loop feedback to accurately estimate the state of the robot with respect to the environment. We used Challenge 2 of the 2017 Mohammed Bin Zayed International Robotics Competition as an example outdoor manipulation task, demonstrating the capabilities of our robot system and approach in achieving reliable performance in the fields, and ranked fifth place internationally in the competition.

show abstract

Section: Related Workmentioning

confidence: 99%

Multimodal sensing and active continuous closed‐loop feedback for achieving reliable manipulation in the outdoor physical world

Chan¹,

Mizohana²,

Chen³

et al. 2018

Journal of Field Robotics

View full text Add to dashboard Cite

show abstract

“…Carrera et al [10] developed a parametric MPs model based on a mixture of several DMPs. First, they record multiple demonstrations.…”

Section: Related Workmentioning

confidence: 99%

Model selection for incremental learning of generalizable movement primitives

Hazara

Kyrki

2017

2017 18th International Conference on Advanced Robotics (ICAR)

View full text Add to dashboard Cite

Although motor primitives (MPs) have been studied extensively, much less attention has been devoted to studying their generalization to new situations. To cope with varying conditions, a MP's policy encoding must support generalization over task parameters to avoid learning separate primitives for each condition. Local and linear parameterized models have been proposed to interpolate over task parameters to provide limited generalization.In this paper, we present a global parametric motion primitive which allows generalization beyond local or linear models. Primitives are modelled using a linear basis function model with global non-linear basis functions. Using the global parametric model, we developed an online incremental learning framework for constructing a database of MPs from a single human demonstration. Above all, we propose a model selection method that can choose an optimal model complexity even with few training samples, which makes it suitable for online incremental learning. Experiments with a ball-in-a-cup task with varying string lengths demonstrate that the global parametric approach can successfully extract underlying regularities in a database of MPs leading to enhanced generalization capability of the parametric MPs and increased speed (convergence rate) of learning. Furthermore, it significantly excels over locally weighted regression both in terms of inter-and extrapolation.

show abstract

“…Carrera et al [6] developed a parametric MPs model based on a mixture of several DMPs. First, they record multiple demonstrations.…”

Section: Related Workmentioning

confidence: 99%

“…where w k denotes the weight vector of a non-parametric DMP optimized for parameter values l k . The initial weights can be merely imitated from a human demonstration using (6) or improved using a policy search method [18]. In either case, reproducing an imitated task using w k should lead to a successful performance in an environment parametrized by l k .…”

Section: Global Parametric Dynamic Movement Primitivesmentioning

confidence: 99%

See 1 more Smart Citation

Generalizing Movement Primitives to New Situations

Lundell

Hazara

Kyrki

2017

Towards Autonomous Robotic Systems

View full text Add to dashboard Cite

Although motor primitives (MPs) have been studied extensively, much less attention has been devoted to studying their generalization to new situations. To cope with varying conditions, a MP's policy encoding must support generalization over task parameters to avoid learning separate primitives for each condition. Local and linear parameterized models have been proposed to interpolate over task parameters to provide limited generalization. In this paper, we present a global parametric motion primitive (GPDMP) which allows generalization beyond local or linear models. Primitives are modeled using a linear basis function model with global non-linear basis functions. The model is constructed from initial non-parametric primitives found using a single human demonstration and subsequent episodes of reinforcement learning to adapt the demonstrated skill to other task parameters. The initial models are then used to optimize the parameters of the global parametric model. Experiments with a ball-in-a-cup task with varying string lengths show that GPDMP allows greatly improved extrapolation compared to earlier local or linear models.

show abstract

Learning multiple strategies to perform a valve turning with underwater currents using an I-AUV

Cited by 18 publications

References 17 publications

Multimodal sensing and active continuous closed‐loop feedback for achieving reliable manipulation in the outdoor physical world

Multimodal sensing and active continuous closed‐loop feedback for achieving reliable manipulation in the outdoor physical world

Model selection for incremental learning of generalizable movement primitives

Generalizing Movement Primitives to New Situations

Contact Info

Product

Resources

About