检索结果-内蒙古大学图书馆

Knowledge-aware Graph Transformer for Pedestrian Trajectory Prediction

学校读者我要写书评

暂无评论

Knowledge-aware Graph Transformer for Pedestrian Trajectory ...

International Conference on Intelligent Transportation

作者： Yu Liu Yuexin Zhang Kunming Li Yongliang Qiao Stewart Worrall You-Fu Li He Kong Shenzhen Key Laboratory of Control Theory and Intelligent Systems Southern University of Science and Technology (SUSTech) Shenzhen China Department of Mechanical Engineering City University of Hong Kong Hong Kong SAR China School of Automation Guangdong Polytechnic Normal University Guangzhou China Australian Centre for Field Robotics The University of Sydney NSW Australia Australian Institute for Machine Learning The University of Adelaide SA Australia Department of Mechanical Engineering City University of Hong Kong Hong Kong SAR China Shenzhen Key Laboratory of Control Theory and Intelligent Systems SUSTech Shenzhen China Guangdong Provincial Key Laboratory of Human-Augmentation Rehabilitation Robotics in Universities SUSTech Shenzhen China

Predicting pedestrian motion trajectories is crucial for path planning and motion control of autonomous vehicles. Accurately forecasting crowd trajectories is challenging due to the uncertain nature of human motions in different environments. For training, recent deep learning-based prediction approaches mainly utilize information like trajectory history and interactions between pedestrians, among others. This can limit the prediction performance across various scenarios since the discrepancies between training datasets have not been properly incorporated. To overcome this limitation, this paper proposes a graph transformer structure to improve prediction performance, capturing the differences between the various sites and scenarios contained in the datasets. In particular, a self-attention mechanism and a domain adaption module have been designed to improve the generalization ability of the model. Moreover, an additional metric considering cross-dataset sequences is introduced for training and performance evaluation purposes. The proposed framework is validated and compared against existing methods using popular public datasets, i.e., ETH and UCY. Experimental results demonstrate the improved performance of our proposed scheme.

关键词：

Biomimetic robots promote the 3Rs Principle in animal testing

学校读者我要写书评

暂无评论

Biomimetic robots promote the 3Rs Principle in animal testin...

2021 Conference on Artificial Life, ALIFE 2021

作者： Bierbach, David Francisco, Fritz Lukas, Juliane Landgraf, Tim Maxeiner, Moritz Romanczuk, Pawel Musiolek, Lea Hafner, Verena V. Krause, Jens Faculty of Life Sciences Division of Biology and Ecology of Fishes Humboldt Universität zu Berlin Berlin Germany Department of Biology and Ecology of Fishes Leibniz Institute of Freshwater Ecology and Inland Fisheries Berlin Germany Dahlem Center for Machine Learning and Robotics Freie Universität Berlin Berlin Germany Faculty of Life Sciences Institute for Theoretical Biology Humboldt Universität zu Berlin Berlin Germany Adaptive Systems Group Department of Computer Science Humboldt-Universität zu Berlin Berlin Germany

In order to strengthen animal welfare, many countries require that experimenters follow the ‘3Rs Principle’ when designing animal experiments. The 3Rs call for a reduction in the number of animals used, the refinement of methods to reduce stress as well as the full replacement of animals in experimentation through alternative methods. Biomimetic robots that resemble live animals and allow for natural-like interactions represent a valuable tool to achieve the 3Rs’ objectives. On the basis of our research with a robotic fish that is accepted as a conspecific by live poeciliid fishes, we highlight how biomimetic robots can reduce the number of animals tested by (a) substituting live animals, (b) providing highly standardized cues, and (c) reducing overall stress for live animals during tests through less handling. © 2021 ALIFE. All Rights Reserved.

关键词： Invertebrates

Asynchronous multi agent active search

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Ghods, Ramina Banerjee, Arundhati Schneider, Jeff School of Computer Science Carnegie Mellon University Robotics Institute School of Computer Science Carnegie Mellon University Department of Machine Learning

Active search refers to the problem of efficiently locating targets in an unknown environment by actively making data-collection decisions, and has many applications including detecting gas leaks, radiation sources or human survivors of disasters using aerial and/or ground robots (agents). Existing active search methods are in general only amenable to a single agent, or if they extend to multi agent they require a central control system to coordinate the actions of all agents. However, such control systems are often impractical in robotics applications. In this paper, we propose two distinct active search algorithms called SPATS (Sparse Parallel Asynchronous Thompson Sampling) and LATSI (LAplace Thompson Sampling with Information gain) that allow for multiple agents to independently make data-collection decisions without a central coordinator. Throughout we consider that targets are sparsely located around the environment in keeping with compressive sensing assumptions and its applicability in real world scenarios. Additionally, while most common search algorithms assume that agents can sense the entire environment (e.g. compressive sensing) or sense point-wise (e.g. Bayesian Optimization) at all times, we make a realistic assumption that each agent can only sense a contiguous region of space at a time. We provide simulation results as well as theoretical analysis to demonstrate the efficacy of our proposed algorithms. Copyright © 2020, The Authors. All rights reserved.

关键词： Multi agent systems

Cost-Effective Communication in UDN in Indoor and Outdoor Environment via machine learning

学校读者我要写书评

暂无评论

Cost-Effective Communication in UDN in Indoor and Outdoor En...

Artificial Intelligence and Knowledge Discovery in Concurrent Engineering (ICECONF), International Conference on

作者： K Nattar Karman V. Velmurugan Kommisetti Murthy Raju T. Sajana V. Vijayalakshmi JoshuvaArockia Dhanraj Departmeru of Artificial Intelligence and Machine Learning Saveetha School of Engineering Chennai Tamil Nadu India Department of Electronics and Communication Engineering Vel Tech Rangarajan and Dr.Sagunthala R&D Institute of science and Technology Chennai Tamil Nadu India Department of Electronics and Communication Engineering Shri Vishnu Engineering College for Women West Godavari Andhra Pradesh India Department of Artificial Intelligence and Data Science KoneruLakshmaiah Education Foundation Vaddeswaram Andhra Pradesh India Department of Networking and Communications School of Computing SRM Institute of Science and Technology Kattankulathur Tamil Nadu India Department of Mechatronics Engineering Centre for Automation and Robotics (ANRO) Hindustan Institute of Technology and Science Chennai Tamil Nadu India

In general, applications on a densely populated network are slower. When there are no opportunities to interact with the devices on the network, the user is forced to communicate at some cost. Thus, the inconsistency that exists there increases the cost over that dense network. The single way to evaluate prospect usage is to look at that have used in the earlier period. This paper proposes an upgraded approach to upgrading existing services within the Ultra Dense cloud Network (UDN). In this way the cost for the services available there is calculated in advance and a system is developed to manage it effectively. This proposed method clearly shows that the effective cost can be managed by combining high-costservices and low-costservices separately. This will help to manage the various communication processes to achieve high effectiveness in Ultra Dense cloud Network (UDN). The proposed method achieves 93.28% of Cost-effective Data Management.49% of Application log failure, 42 % of system log failure, 98% of network log management, 96% of security log management, 99% cost effective management.

关键词： Costs Mobility models machine learning Knowledge discovery Scheduling Energy efficiency Concurrent engineering

Neural dynamical systems: Balancing structure and flexibility in physical prediction

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Mehta, Viraj Char, Ian Neiswanger, Willie Chung, Youngseog Nelson, Andrew Oakleigh Boyer, Mark D. Kolemen, Egemen Schneider, Jeff Robotics Institute Carnegie Mellon University United States Machine Learning Department Carnegie Mellon University United States Princeton Plasma Physics Laboratory United States

We introduce Neural Dynamical Systems (NDS), a method of learning dynamical models in various gray-box settings which incorporates prior knowledge in the form of systems of ordinary differential equations. NDS uses neural networks to estimate free parameters of the system, predicts residual terms, and numerically integrates over time to predict future states. A key insight is that many real dynamical systems of interest are hard to model because the dynamics may vary across rollouts. We mitigate this problem by taking a trajectory of prior states as the input to NDS and train it to dynamically estimate system parameters using the preceding trajectory. We find that NDS learns dynamics with higher accuracy and fewer samples than a variety of deep learning methods that do not incorporate the prior knowledge and methods from the system identification literature which do. We demonstrate these advantages first on synthetic dynamical systems and then on real data captured from deuterium shots from a nuclear fusion reactor. Finally, we demonstrate that these benefits can be utilized for control in small-scale experiments. Copyright © 2020, The Authors. All rights reserved.

关键词： Dynamical systems

Molecular Contrastive learning of Representations via Graph Neural Networks

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Wang, Yuyang Wang, Jianren Cao, Zhonglin Farimani, Amir Barati Department of Mechanical Engineering Carnegie Mellon University PittsburghPA15213 United States Machine Learning Department Carnegie Mellon University PittsburghPA15213 United States Robotics Institute Carnegie Mellon University PittsburghPA15213 United States Department of Chemical Engineering Carnegie Mellon University PittsburghPA15213 United States

Molecular machine learning (ML) bears promise for efficient molecule property prediction and drug discovery. However, labeled molecule data can be expensive and time-consuming to acquire. Due to the limited labeled data, it is a great challenge for supervised-learning ML models to generalize to the giant chemical space. In this work, we present MolCLR: Molecular Contrastive learning of Representations via Graph Neural Networks (GNNs), a self-supervised learning framework that leverages large unlabeled data (∼10M unique molecules). In MolCLR pre-training, we build molecule graphs and develop GNN encoders to learn differentiable representations. Three molecule graph augmentations are proposed: atom masking, bond deletion, and subgraph removal. A contrastive estimator maximizes the agreement of augmentations from the same molecule while minimizing the agreement of different molecules. Experiments show that our contrastive learning framework significantly improves the performance of GNNs on various molecular property benchmarks including both classification and regression tasks. Benefiting from pre-training on the large unlabeled database, MolCLR even achieves state-of-the-art on several challenging benchmarks after fine-tuning. Additionally, further investigations demonstrate that MolCLR learns to embed molecules into representations that can distinguish chemically reasonable molecular similarities. Copyright © 2021, The Authors. All rights reserved.

关键词： Molecules

Graph network for simultaneous learning of forward and inverse physics

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Prakash, Sakthi Kumar Arul Tucker, Conrad Department of Mechanical Engineering Carnegie Mellon University United States Department of Machine Learning Carnegie Mellon University United States The Robotics Institute Carnegie Mellon University United States CyLab Security and Privacy Institute Carnegie Mellon University United States Department of Biomedical Engineering Carnegie Mellon University 5000 Forbes Avenue PittsburghPA15213 United States

In this work, we propose an end-to-end graph network that learns forward and inverse models of particle-based physics using interpretable inductive biases. Physics-informed neural networks are often engineered to solve specific problems through problem-specific regularization and loss functions. Such explicit learning biases the network to learn data specific patterns and may require a change in the loss function or neural network architecture hereby limiting their generalizabiliy. While recent studies have proposed graph networks to study forward dynamics, they rely on particle specific parameters such as mass, etc. to approximate the dynamics of the system. Our graph network is implicitly biased by learning to solve several tasks, thereby sharing representations between tasks in order to learn the forward dynamics as well as infer the probability distribution of unknown particle specific properties. We evaluate our approach on one-step next state prediction tasks across diverse datasets that feature different particle interactions. Our comparison against related data-driven physics learning approaches reveals that our model is able to predict the forward dynamics with at least an order of magnitude higher accuracy. We also show that our approach is able to recover multi-modal probability distributions of unknown physical parameters using orders of magnitude fewer samples. Copyright © 2021, The Authors. All rights reserved.

关键词： Probability distributions

Shared control of robot-robot collaborative lifting with agent postural and force ergonomic optimization

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Rapetti, Lorenzo Tirupachuri, Yeshasvi Ranavolo, Alberto Kawakami, Tomohiro Yoshiike, Takahide Pucci, Daniele Dynamic Interaction Control at Istituto Italiano di Tecnologia Center for Robotics Technologies Genova Italy Machine Learning and Optimisation University of Manchester Manchester United Kingdom Department of Occupational and Environmental Medicine Epidemiology and Hygiene INAIL Roma Italy Honda R&D Co. Ltd. Saitama Japan

Humans show specialized strategies for efficient collaboration. Transferring similar strategies to humanoid robots can improve their capability to interact with other agents, leading the way to complex collaborative scenarios with multiple agents acting on a shared environment. In this paper we present a control framework for robot-robot collaborative lifting. The proposed shared controller takes into account the joint action of both the robots thanks to a centralized controller that communicates with them, and solves the whole-system optimization. Efficient collaboration is ensured by taking into account the ergonomic requirements of the robots through the optimization of posture and contact forces. The framework is validated in an experimental scenario with two iCub humanoid robots performing different payload lifting sequences. Copyright © 2021, The Authors. All rights reserved.

关键词： Ergonomics

Enabling robot selective trained deep neural networks for object detection through intelligent infrastructure 2019

学校读者我要写书评

暂无评论

Enabling robot selective trained deep neural networks for ob...

4th International Conference on Automation, Control and robotics Engineering, CACRE 2019

作者： Poss, Christian Irrenhauser, Thomas Prueglmeier, Marco Goehring, Daniel Zoghlami, Firas Salehi, Vahid Ibragimov, Olimjon Logistics Robotics BMW Group Munich Germany Dahlem Center for Robotics and Machine Learning Freie Universität Berlin Germany Department of Applied Sciences and Mechatronic University of Applied Sciences Munich Germany School of International Business and Entrepreneurship Steinbeis University Berlin Germany

ISBN: (纸本)9781450371865

To save costs in logistics, handling steps are going to be automated by robots in the future. Due to the complex industrial conditions prevailing there, this is only possible with a sufficient degree of intelligence in the respective systems. Despite advances in artificially intelligent algorithms, the latest neural networks reveal significant weaknesses in performance and transferability to other applications. In order to enable a holistic autonomous material flow, the paper presents an infrastructure concept, which makes it possible to identify and train the most suitable networks robot-selectively with very limited effort. Using two practical examples, the functionality of the designed algorithms for the industrial implementation of a new use case as well as the updating and improvement of an existing system is finally outlined. It will be shown that with measures such as the automated collection of training data, the AI-supported labeling process, the intuitive validation of the trained networks via a mobile application and the automated retraining of robots already integrated, a further step can be taken towards holistically automated logistics process chains. © 2019 Association for Computing machinery.

关键词： Object detection