检索结果-内蒙古大学图书馆

Attention in Cognitive systems. Theories and systems from an Interdisciplinary Viewpoint 1

学校读者我要写书评

暂无评论

丛书名： Lecture Notes in Computer Science

1000年

作者： Lucas Paletta Erich Rome

ISBN: (数字)9783540773436

ISBN: (纸本)9783540773429

Attentionhasbeenrepresentingacorescienti?ctopicinthedesignofAI-enabled systems within the last decades. Today, in the ongoing debate, design, and c- putationalmodelingofarti?cialcognitivesystems,attentionhasgainedacentral position as a focus of research. For instance, attentional methods are considered in investigating the interfacing of sensory and cognitive information processing, for the organization of behaviors, and for the understanding of individual and social cognition in re?ection of infant development. Whilevisualcognitionplaysacentralroleinhumanperception,?ndingsfrom neuroscience and experimental psychology have provided strong evidence about ***-motor intelligence requires a continuous and focused interplay between the control of motor activities and the interpretation of feedback from perceptual modalities. Decision making about the selection of information from the incoming sensory stream – in tune with contextual processing on a current task and an agent’s global objectives – becomes a further challenging issue in attentional control. Attention must operate at interfaces between bottom-up driven world int- pretation and top-down driven information selection, thus acting at the core of arti?cial cognitive systems. These insights have already induced changes in AI-related disciplines, such as the design of behavior-based robot control and the computational modeling of animats. Today, the development of enabling technologiessuch as autonomous robotic systems,miniaturizedmobile–evenwearable–sensors,andambientintelligence systems involves the real-time analysis of enormous quantities of data. These data have to be processed in an intelligent way to provide “on time delivery”of the required relevant information. Knowledge has to be applied about what needs to be attended to, and when, and what to do in a meaningful sequence, in correspondence with visual feedback.

关键词： Theory of Computation Artificial Intelligence Image Processing and Computer Vision Pattern Recognition Computer Graphics Neurosciences

A two-level adaptive target recognition and tracking method based on vision for multi-robot system

学校读者我要写书评

暂无评论

A two-level adaptive target recognition and tracking method ...

2019 IEEE International Conference on Robotics and Biomimetics, ROBIO 2019

作者： Ren, Liang Cao, Zhiqiang Tan, Min Zhao, Peng Chen, Xuechao University of Chinese Academy of Sciences State Key Laboratory of Management and Control for Complex Systems CASIA Beijing China 21st Research Institute of China Electronics Technology Group Corporation Shanghai China BIT Beijing Advanced Innovation Center for Intelligent Robots and Systems Beijing China

ISBN: (纸本)9781728163215

The vision-based target recognition and tracking have received much attention in the field of robotics. Existing methods mainly focus on the vision perception of individual robot with a single view, however, the performance is susceptible to illumination and occlusion. Multi-robot collaborative perception provides a potential solution to deal with the limitation of single-view observation, however, the challenging of environmental adaptability for multi-robot collaborative decision still remains unsolved. To solve this problem, this paper proposes a two-level adaptive target recognition and tracking method based on vision for multi-robot system. The problem of multi-robot target recognition and tracking is solved under a two-level framework, which contains the features fusion level of individual robot and the cooperation level of multi-robot system. In the first level, the features measuring results that influence the visual perception of individual robot are fused, while the second level combines the voting of each robot together to determine the target for multi-robot system. Both the features measuring weights and robots voting weights are adaptively updated according to their evaluation, which lead to a beneficial result where the features and robots with higher accuracy play major roles in the first and second levels, respectively. Therefore, a good adaptability to the environments can be guaranteed. The experimental results show that the proposed approach can realize the coordination of multi-robot system in target recognition and tracking with an effective performance. © 2019 IEEE.

关键词： Multipurpose robots

MoGaze: A dataset of full-body motions that includes workspace geometry and eye-gaze

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Kratzer, Philipp Bihlmaier, Simon Midlagajni, Niteesh Balachandra Prakash, Rohit Toussaint, Marc Mainprice, Jim Machine Learning and Robotics Lab University of Stuttgart Humans to Robots Motions Research Group University of Stuttgart Germany Humans to Robots Motions Research Group University of Stuttgart Germany Learning and Intelligent Systems lab TU Berlin Germany

As robots become more present in open human environments, it will become crucial for robotic systems to understand and predict human motion. Such capabilities depend heavily on the quality and availability of motion capture data. However, existing datasets of full-body motion rarely include 1) long sequences of manipulation tasks, 2) the 3D model of the workspace geometry, and 3) eye-gaze, which are all important when a robot needs to predict the movements of humans in close proximity. Hence, in this paper, we present a novel dataset of full-body motion for everyday manipulation tasks, which includes the above. The motion data was captured using a traditional motion capture system based on reflective markers. We additionally captured eye-gaze using a wearable pupil-tracking device. As we show in experiments, the dataset can be used for the design and evaluation of full-body motion prediction algorithms. Furthermore, our experiments show eye-gaze as a powerful predictor of human intent. The dataset includes 180 min of motion capture data with 1627 pick and place actions being performed. It is available at MoGaze Dataset1 and is planned to be extended to collaborative tasks with two humans in the near future. Copyright © 2020, The Authors. All rights reserved.

关键词： Motion estimation

Grasping of Planar Objects Using Visual Perception

学校读者我要写书评

暂无评论

Grasping of Planar Objects Using Visual Perception

IEEE-RAS International Conference on Humanoid robots

作者： Madjid Boudaba Alicia Casals Design Center TES Electronic Solution GmbH Stuttgart Germany GRINS: Research Group on Intelligent Robots and Systems Technical University of Catalonia Barcelona Spain

ISBN: (纸本)1424401992

This paper presents an algorithm to compute a feasible region for grasping unknown 2D objects using visual features. The algorithm is divided into two sections: visual information extraction that contains the object's features and grasp planning that contains the valid grasps. A relationship between visual features and grasp planning is proposed. Our approach uses three-fingers contact with friction and defines valid and robust grasps from a set of candidate grasps. Each set of three candidate of grasping points is formulated as linear constraints and solved using linear programming solvers. An experimental environment was developed to test this method on three objects.

关键词： Visual perception Grasping Friction Tactile sensors Fingers robots Force control Strain control Solids Shape control

Automated generation of reactive programs from human demonstration for orchestration of robot behaviors

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Berenz, Vincent Bjelic, Ahmed Mainprice, Jim Autonomous Motion Department Max Planck Institute for Intelligent Systems Tübingen Germany Humans to Robots Motion Research Group HRM Stuttgart University Stuttgart Germany

Social robots or collaborative robots that have to interact with people in a reactive way are difficult to program. This difficulty stems from the different skills required by the programmer: To provide an engaging user experience the behavior must include a sense of aesthetics while robustly operating in a continuously changing environment. The Playful framework [1] allows to compose such dynamic behaviors using a basic set of action and perception primitives. Within this framework, a behavior is encoded as a list of declarative statements corresponding to high level sensory-motor couplings. To facilitate non expert users to program such behaviors, we propose a Learning from Demonstration (LfD) technique that maps motion capture of humans directly to a Playful script. The approach proceeds by identifying the sensory-motor couplings that are active at each step using the Viterbi path in a Hidden Markov Model (HMM). Given these activation patterns, binary classifiers called evaluations are trained to associate activations to sensory data. Modularity is increased by clustering the sensory-motor couplings, leading to a hierarchical tree structure. The novelty of the proposed approach is that the learned behavior is encoded not in terms of trajectories in a task space, but as couplings between sensory information and high level motor actions. This provides advantages in terms of behavioral generalization and reactivity displayed by the robot. Copyright © 2019, The Authors. All rights reserved.

关键词： Couplings

Bonn activity maps: Dataset description

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Tanke, Julian Kwon, Oh-Hun Stotko, Patrick Rosu, Radu A. Weinmann, Michael Errami, Hassan Behnke, Sven Bennewitz, Maren Klein, Reinhard Weber, Andreas Yao, Angela Gall, Juergen Computer Vision Group University of Bonn Computer Graphics Group University of Bonn Autonomous Intelligent Systems University of Bonn Virtual Reality Group University of Bonn Humanoid Robots Lab University of Bonn National University of Singapore

The key prerequisite for accessing the huge potential of current machine learning techniques is the availability of large databases that capture the complex relations of interest. Previous datasets are focused on either 3D scene representations with semantic information, tracking of multiple persons and recognition of their actions, or activity recognition of a single person in captured 3D environments. We present Bonn Activity Maps, a large-scale dataset for human tracking, activity recognition and anticipation of multiple persons. Our dataset comprises four different scenes that have been recorded by time-synchronized cameras each only capturing the scene partially, the reconstructed 3D models with semantic annotations, motion trajectories for individual people including 3D human poses as well as human activity annotations. We utilize the annotations to generate activity likelihoods on the 3D models called activity maps. Copyright © 2019, The Authors. All rights reserved.

关键词： Motion estimation

Planning Coordinated Human-Robot Motions with Neural Network Full-Body Prediction Models

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Kratzer, Philipp Toussaint, Marc Mainprice, Jim The Machine Learning and Robotics Lab University of Stuttgart Germany The Humans to Robots Motions Research Group University of Stuttgart Germany The Learning and Intelligent Systems Lab TU Berlin Germany

Numerical optimization has become a popular approach to plan smooth motion trajectories for robots. However, when sharing space with humans, balancing properly safety, comfort and efficiency still remains challenging. This is notably the case because humans adapt their behavior to that of the robot, raising the need for intricate planning and prediction. In this paper, we propose a novel optimization-based motion planning algorithm, which generates robot motions, while simultaneously maximizing the human trajectory likelihood under a data-driven predictive model. Considering planning and prediction together allows us to formulate objective and constraint functions in the joint human-robot state space. Key to the approach are added latent space modifiers to a differentiable human predictive model based on a dedicated recurrent neural network. These modifiers allow to change the human prediction within motion optimization. We empirically evaluate our method using the publicly available MoGaze [1] dataset. Our results indicate that the proposed framework outperforms current baselines for planning handover trajectories and avoiding collisions between a robot and a human. Our experiments demonstrate collaborative motion trajectories, where both, the human prediction and the robot plan, adapt to each other. © 2022, CC BY.

关键词： Forecasting

Design and Research on a Vision-based Lightweight Soft Ankle Exosuit 8

学校读者我要写书评

暂无评论

Design and Research on a Vision-based Lightweight Soft Ankle...

8th IEEE International Conference on Advanced Robotics and Mechatronics, ICARM 2023

作者： Bao, Bingsheng Tu, Yao Dang, Diyang Zhu, Aibin Wang, Xin Zhou, Xu Gao, Chong Zhang, Zheng Xu, Haibo Institute of Robotics & Intelligent Systems Xi'An Jiaotong University Shaanxi Key Laboratory of Intelligent Robots Xi'an710049 China Xi'An Jiaotong University Xi'an710049 China School of Construction Machinery Chang'An University Xi'an710064 China School of Mechanical Engineering Northwestern Polytechnical University Xi'an710072 China Taiyuan Research Institute of China Coal Technology & Engineering Group Co. Ltd Taiyuan030006 China

ISBN: (纸本)9798350300178

There has been growing interest in the development of exosuit-assisted locomotion to effectively assist human motion. In this study, we present a novel soft ankle joint exosuit capable of providing dynamic support in varying terrains. The exosuit's underdriven structural design, combined with the use of visual sensors for real-time environmental detection, distinguishes it from existing solutions. The underdriven design enables the exosuit to achieve two degrees of freedom using a single actuator, thus improving its mobility and functionality. Additionally, we employ a semantic segmentation model based on deep learning to recognize the terrain of the environment and adapt the exosuit's assistance accordingly. The visual sensors provide real-time information to the system, allowing it to switch between assisted states and improve the exosuit's overall performance. We carried out experiments on a prototype to validate our proposed approach. With 83% accuracy of terrain recognition, the results demonstrate its feasibility and potential for future development. © 2023 IEEE.

关键词： Structural design

Anticipating Human Intention for Full-Body Motion Prediction in Object Grasping and Placing Tasks

学校读者我要写书评

暂无评论

Anticipating Human Intention for Full-Body Motion Prediction...

IEEE International Workshop on Robot and Human Communication (ROMAN)

作者： Philipp Kratzer Niteesh Balachandra Midlagajni Marc Toussaint Jim Mainprice Machine Learning and Robotics Lab University of Stuttgart Germany Humans to Robots Motions Research Group HRM University of Stuttgart Germany Learning and Intelligent Systems Lab Technical University of Berlin Germany

ISBN: (数字)9781728160757

ISBN: (纸本)9781728160764

Motion prediction in unstructured environments is a difficult problem and is essential for safe and efficient human-robot space sharing and collaboration. In this work, we focus on manipulation movements in environments such as homes, workplaces or restaurants, where the overall task and environment can be leveraged to produce accurate motion prediction. For these cases we propose an algorithmic framework that accounts explicitly for the environment geometry based on a model of affordances and a model of short-term human dynamics both trained on motion capture data. We propose dedicated function networks for graspability and placebility affordances and we make use of a dedicated RNN [1] for short-term motion prediction. The prediction of grasp and placement probability densities are used by a constraint-based trajectory optimizer to produce a full-body motion prediction over the entire horizon. We show by comparing to ground truth data that we achieve similar performance for full-body motion predictions as using oracle grasp and place locations.

关键词：