检索结果-内蒙古大学图书馆

Prediction of human full-body movements with motion optimization and recurrent neural networks

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Kratzer, Philipp Toussaint, Marc Mainprice, Jim Machine Learning and Robotics Lab University of Stuttgart Germany Max Planck Institute for Intelligent Systems IS-MPI Tübingen Germany

Human movement prediction is difficult as humans naturally exhibit complex behaviors that can change drastically from one environment to the next. In order to alleviate this issue, we propose a prediction framework that decouples short-Term prediction, linked to internal body dynamics, and long-Term prediction, linked to the environment and task constraints. In this work we investigate encoding short-Term dynamics in a recurrent neural network, while we account for environmental constraints, such as obstacle avoidance, using gradient-based trajectory optimization. Experiments on real motion data demonstrate that our framework improves the prediction with respect to state-of-The-Art motion prediction methods, as it accounts to beforehand unseen environmental structures. Moreover we demonstrate on an example, how this framework can be used to plan robot trajectories that are optimized to coordinate with a human partner. Copyright © 2019, The Authors. All rights reserved.

关键词： Motion estimation

Deep workpiece region segmentation for bin picking

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Khalid, Muhammad Usman Hager, Janik M. Kraus, Werner Huber, Marco F. Toussaint, Marc Robot and Assistive Systems Fraunhofer IPA Stuttgart Germany Machine Learning & Robotics Lab University of Stuttgart Germany Fraunhofer IPA Stuttgart and Institute of Industrial Manufacturing and Management IFF University of Stuttgart Germany

For most industrial bin picking solutions, the pose of a workpiece is localized by matching a CAD model to point cloud obtained from 3D sensor. Distinguishing flat workpieces from bottom of the bin in point cloud imposes challenges in the localization of workpieces that lead to wrong or phantom detections. In this paper, we propose a framework that solves this problem by automatically segmenting workpiece regions from non-workpiece regions in a point cloud data. It is done in real time by applying a fully convolutional neural network trained on both simulated and real data. The real data has been labelled by our novel technique which automatically generates ground truth labels for real point clouds. Along with real time workpiece segmentation, our framework also helps in improving the number of detected workpieces and estimating the correct object poses. Moreover, it decreases the computation time by approximately 1s due to a reduction of the search space for the object pose estimation. Copyright © 2019, The Authors. All rights reserved.

关键词： Chemical detection

Trajectory-based off-policy deep reinforcement learning

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Doerr, Andreas Volpp, Michael Toussaint, Marc Trimpe, Sebastian Daniel, Christian Bosch Center for Artificial Intelligence Renningen Germany Max Planck Institute for Intelligent Systems Stuttgart/Tubingen Germany Machine Learning and Robotics Lab University of Stuttgart Germany

Policy gradient methods are powerful reinforcement learning algorithms and have been demonstrated to solve many complex tasks. However, these methods are also data-inefficient, afflicted with high variance gradient estimates, and frequently get stuck in local optima. This work addresses these weaknesses by combining recent improvements in the reuse of off-policy data and exploration in parameter space with deterministic behavioral policies. The resulting objective is amenable to standard neural network optimization strategies like stochastic gradient descent or stochastic gradient Hamiltonian Monte Carlo. Incorporation of previous rollouts via importance sampling greatly improves data-efficiency, whilst stochastic optimization schemes facilitate the escape from local optima. We evaluate the proposed approach on a series of continuous control benchmark tasks. The results show that the proposed algorithm is able to successfully and reliably learn solutions using fewer system interactions than standard policy gradient methods. Copyright © 2019, The Authors. All rights reserved.

关键词： Deep learning

Bayesian Model Selection of Lithium-Ion Battery Models via Bayesian Quadrature

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Adachi, Masaki Kuhn, Yannick Horstmann, Birger Osborne, Michael A. Howey, David A. Machine Learning Research Group University of Oxford OX2 6ED United Kingdom Battery Intelligence Lab University of Oxford OX1 3PJ United Kingdom Pfaffenwaldring 38-40 Stuttgart70569 Germany Helmholtz Institute Ulm Helmholtzstraße 11 Ulm89081 Germany Universität Ulm Albert-Einstein-Allee 47 Ulm89081 Germany The Faraday Institution Harwell Campus DidcotOX11 0RA United Kingdom

This paper presents a Bayesian model selection approach via Bayesian quadrature and sensitivity analysis of the selection criterion for a lithium-ion battery model. The Bayesian model evidence is adopted as the metric, which can select the simplest but well-describing model based on Occam's razor principle. While the model evidence requires prohibitive integral computations over parameter space, Bayesian quadrature offers sample-efficient integration via model-based inference to minimise the number of battery model evaluations. The posterior distribution of battery model parameters can also be inferred as a byproduct in one go, which is also beneficial in creating a digital twin. The simplest lithium-ion battery models, equivalent circuit models, were used to analyse the sensitivity of the selection criterion at given different datasets and model configurations. We show that popular selection criteria, such as root-mean-square error, and Bayesian information criterion, can fail to select a correct model in a multimodal posterior case. The model evidence can spot the true model in such cases, simultaneously providing the variance of evidence inference itself as an indication of confidence. Bayesian quadrature can compute the evidence faster than popular MCMC *** Codes 62C10, 62F15 © 2022, CC BY-NC-ND.

关键词： Parameter estimation

Prior Visual Relationship Reasoning For Visual Question Answering

学校读者我要写书评

暂无评论

Prior Visual Relationship Reasoning For Visual Question Answ...

IEEE International Conference on Image Processing

作者： Zhuoqian Yang Zengchang Qin Jing Yu Tao Wan Robotics Institute Carnegie Mellon University Pittsburgh PA USA Intelligent Computing and Machine Learning Lab School of ASEE Beihang University China Institute of Information Engineering CAS China School of Biological Science and Medical Engineering Beihang University Beijing China

ISBN: (数字)9781728163956

ISBN: (纸本)9781728163963

Visual Question Answering (VQA) is a representative task of cross-modal reasoning where an image and a free-form question in natural language are presented and the correct answer needs to be determined using both visual and textual information. One of the key issues of VQA is to reason with semantic clues in the visual content under the guidance of the question. In this paper, we propose Scene Graph Convolutional Network (SceneGCN) to jointly reason the object properties and their semantic relations for the correct answer. The visual relationship is projected into a deep learned semantic space constrained by visual context and language priors. Based on comprehensive experiments on two challenging datasets: GQA and VQA 2.0, we demonstrate the effectiveness and interpretability of the new model.

关键词： Visualization Semantics Convolution Cognition Knowledge discovery Benchmark testing Measurement

Kinematic morphing networks for manipulation skill transfer

学校读者我要写书评

暂无评论

arXiv 2018年

作者： Englert, Peter Toussaint, Marc Machine Learning & Robotics Lab University of Stuttgart Germany

The transfer of a robot skill between different geometric environments is non-trivial since a wide variety of environments exists, sensor observations as well as robot motions are high-dimensional, and the environment might only be partially observed. We consider the problem of extracting a low-dimensional description of the manipulated environment in form of a kinematic model. This allows us to transfer a skill by defining a policy on a prototype model and morphing the observed environment to this prototype. A deep neural network is used to map depth image observations of the environment to morphing parameter, which include transformation and configuration parameters of the prototype model. Using the concatenation property of affine transformations and the ability to convert point clouds to depth images allows to apply the network in an iterative manner. The network is trained on data generated in a simulator and on augmented data that is created by using network predictions. The algorithm is evaluated on different tasks, where it is shown that iterative predictions lead to a higher accuracy than one-step predictions. Copyright © 2018, The Authors. All rights reserved.

关键词： Deep neural networks

Constrained Bayesian optimization of combined interaction force/task space controllers for manipulations

学校读者我要写书评

暂无评论

Constrained Bayesian optimization of combined interaction fo...

2017 IEEE International Conference on robotics and Automation, ICRA 2017

作者： Dries, Danny Englert, Peter Toussaint, Marc Machine Learning and Robotics Lab University of Stuttgart Germany

ISBN: (纸本)9781509046331

In this paper, we address the problem of how a robot can optimize parameters of combined interaction force/task space controllers under a success constraint in an active way. To enable the robot to explore its environment robustly, safely and without the risk of damaging anything, suitable control concepts have to be developed that enable compliant and force control in situations that are afflicted with high uncertainties. Instances of such concepts are impedance, operational space or hybrid control. However, the parameters of these controllers have to be tuned precisely in order to achieve reasonable performance, which is inherently challenging, as often no sufficient model of the environment is available. To overcome this, we propose to use constrained Bayesian optimization to enable the robot to tune its controller parameters autonomously. Unlike other controller tuning methods, this method allows us to include a success constraint into the optimization. Further, we introduce novel performance measures for compliant, force controlled robots. In real world experiments we show that our approach is able to optimize the parameters for a task that consists of establishing and maintaining contact between the robot and the environment efficiently and successfully. © 2017 IEEE.

关键词： Constrained optimization

Deep learning Approach for Linear Locomotion Control of Spherical Robot

学校读者我要写书评

暂无评论

Deep Learning Approach for Linear Locomotion Control of Sphe...

International Conference on Control, Automation and Systems ( ICCAS)

作者： Daniel W Nam Chaehyeuk Lee Jaehwan Choi Yeonjun Kim Soon-Geul Lee KC Machine Learning Lab (ML2) Seoul Korea Medipixel Seoul Korea HANCOM Robotics Pangyo Korea Kyung Hee University Yongin Korea

Spherical robot is a typical nonlinear underactuated nonholonomic system that is difficult to control accurate path trajectory and uniform and stable posture during transfer. In this paper, we propose a method called controller guided model learning (CGML) to control its posture and trajectory path simultaneously. This model-based control algorithm is a combination of traditional classical control techniques and model learning through deep neural networks. CGML implemented with artificial neural network improves data efficiency in learning by using reliability advantage of classical controller. The proposed method can reduce the learning time required for convergence, which is suitable for direct online learning. The simulation environment experiment of the spherical robot showed that CGML is superior to other algorithms in terms of data efficiency.

关键词：

learning to Control Redundant Musculoskeletal Systems with Neural Networks and SQP: Exploiting Muscle Properties

学校读者我要写书评

暂无评论

Learning to Control Redundant Musculoskeletal Systems with N...

IEEE International Conference on robotics and Automation

作者： Danny Driess Heiko Zimmermann Simon Wolfen Dan Suissa Daniel Haeufle Daniel Hennes Marc Toussaint Syn Schmitt Machine Learning and Robotics Lab University of Stuttgart Germany Biomechanics and Biorobotics Group University of Stuttgart Germany Multi-Level Modeling in Motor Control and Rehabilitation Robotics University of Tübingen Germany

Modeling biomechanical musculoskeletal systems reveals that the mapping from muscle stimulations to movement dynamics is highly nonlinear and complex, which makes it difficult to control those systems with classical techniques. In this work, we not only investigate whether machine learning approaches are capable of learning a controller for such systems. We are especially interested in the question if the structure of the musculoskeletal apparatus exhibits properties that are favorable for the learning task. In particular, we consider learning a control policy from target positions to muscle stimulations. To account for the high actuator redundancy of biomechanical systems, our approach uses a learned forward model represented by a neural network and sequential quadratic programming to obtain the control policy, which also enables us to alternate the co-contraction level and hence allows to change the stiffness of the system and to include optimality criteria like small muscle stimulations. Experiments on both a simulated musculoskeletal model of a human arm and a real biomimetic muscle-driven robot show that our approach is able to learn an accurate controller despite high redundancy and nonlinearity, while retaining sample efficiency.

关键词： Muscles Joints Robots Biological system modeling Torque Biomechanics