检索结果-内蒙古大学图书馆

GISR: Geometric Initialization and Silhouette-based Refinement for Single-View Robot Pose and Configuration Estimation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Bilić, Ivan Marić, Filip Bonsignorio, Fabio Petrović, Ivan University of Zagreb Faculty of Electrical Engineering and Computing Laboratory for Autonomous Systems and Mobile Robotics Croatia University of Toronto Institute for Aerospace Studies Space and Terrestrial Autonomous Robotic Systems Laboratory Canada

In autonomous robotics, measurement of the robot’s internal state and perception of its environment, including interaction with other agents such as collaborative robots, are essential. Estimating the pose of the robot arm from a single view has the potential to replace classical eye-to-hand calibration approaches and is particularly attractive for online estimation and dynamic environments. In addition to its pose, recovering the robot configuration provides a complete spatial understanding of the observed robot that can be used to anticipate the actions of other agents in advanced robotics use cases. Furthermore, this additional redundancy enables the planning and execution of recovery protocols in case of sensor failures or external disturbances. We introduce GISR - a deep configuration and robot-to-camera pose estimation method that prioritizes execution in real-time. GISR consists of two modules: (i) a geometric initialization module that efficiently computes an approximate robot pose and configuration, and (ii) a deep iterative silhouette-based refinement module that arrives at a final solution in just a few iterations. We evaluate GISR on publicly available data and show that it outperforms existing methods of the same class in terms of both speed and accuracy, and can compete with approaches that rely on ground-truth proprioception and recover only the pose. Our code is available at https://***/iwhitey/GISR-robot. © 2024, CC BY.

关键词：

Towards the Application of Operational Design Domain Based Scene Generation for Artificial Intelligence Training in Railway Automation The project “Rail Automation with Artificial Intelligence for Detection of Exceptional Situations” (RailAIxs) received funding in the mFUND conveyor line by the German Federal Ministry for Digital and Transport under the funding code 19FS2031A-D.

学校读者我要写书评

暂无评论

Towards the Application of Operational Design Domain Based S...

International Conference on Intelligent Transportation

作者： Till Mersmann Friedrich Betz Julian Eichenbaum Fabian Hampel Simon Klamt Yannick Otten Ingrid Scholl Christian Schindler Faculty of Mechanical Engineering Institute for Rail Vehicle and Transport Systems (IFS) RWTH Aachen University Aachen Germany Faculty of Electrical Engineering and Information Technology Mobile Autonomous Systems and Cognitive Robotics Institute (MASCOR) FH Aachen University of Applied Sciences Aachen Germany

ISBN: (数字)9798331505929

ISBN: (纸本)9798331505936

For automated, driverless rail transportation applications in open environments, Artificial Intelligence (AI)-based methods are gaining importance, especially in computer vision and perception tasks. The safe operation of complex automated systems requires validation processes. For this purpose, the concept of Operational Design Domains (ODDs), driven by recent developments in the automotive industry, is gaining momentum, allowing to describe different aspects of operating conditions as scenes and scenarios. With regard to safety and authorization using AI-based vision systems, data coverage is needed, which can be enhanced by employing virtual reality in different forms. The creation of virtual scenes and sensor models allows the generation of synthetic sensor data and metadata that can be used as a database for the training of the vision system.

关键词： Training Solid modeling Automation Machine vision Virtual reality Rail transportation Safety Artificial intelligence Standards Synthetic data

Euclidean Equivariant Models for Generative Graphical Inverse Kinematics

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Limoyo, Oliver Marić, Filip Giamou, Matthew Alexson, Petra Petrović, Ivan Kelly, Jonathan Institute for Aerospace Studies University of Toronto Canada Laboratory for Autonomous Systems and Mobile Robotics University of Zagreb Croatia

Quickly and reliably finding accurate inverse kinematics (IK) solutions remains a challenging problem for robotic manipulation. Existing numerical solvers typically produce a single solution only and rely on local search techniques to minimize a highly nonconvex objective function. Recently, learning-based approaches that approximate the entire feasible set of solutions have shown promise as a means to generate multiple fast and accurate IK results in parallel. However, existing learning-based techniques have a significant drawback: each robot of interest requires a specialized model that must be trained from scratch. To address this shortcoming, we investigate a novel distance-geometric robot representation coupled with a graph structure that allows us to leverage the flexibility of graph neural networks (GNNs). We use this approach to train a generative graphical inverse kinematics solver (GGIK) that is able to produce a large number of diverse solutions in parallel while also generalizing well—a single learned model can be used to produce IK solutions for a variety of different robots. The graphical formulation elegantly exposes the symmetry and Euclidean equivariance of the IK problem that stems from the spatial nature of robot manipulators. We exploit this symmetry by encoding it into the architecture of our learned model, yielding a flexible solver that is able to produce sets of IK solutions for multiple robots. Copyright © 2023, The Authors. All rights reserved.

关键词： Inverse kinematics

Optimization of Image Preprocessing and Background Influences using a Depth Camera for Person Re-Identification on a mobile Robot

学校读者我要写书评

暂无评论

Optimization of Image Preprocessing and Background Influence...

IEEE International Conference on Automation Science and Engineering (CASE)

作者： Sebastian Flores Zeynep Boztoprak Jana Jost Department of AI and Autonomous Systems Fraunhofer Institute for Material Flow and Logistics Dortmund Germany Department of Robotics and Cognitive Systems Fraunhofer Institute for Material Flow and Logistics Dortmund Germany

In this paper, we optimize datasets with different image preprocessing techniques for person re-identification using the ResNet18 model on a mobile robot with limited hardware, e.g., computational power and depth camera. For this, we create 16 datasets for which we discovered that the inverted original images, from the IR gray value images of the depth camera, results in the highest values with an r1, r5 and mAP of 98.48 %,99.82 % and 80.86 %. Additionally, we explore the cross-dataset evaluation for the 16 datasets to examine the robustness of our model, which points to a low generalizability. The scores are associated with the similarity between the trained and evaluated dataset.

关键词：

Auto-Augmentation Contrastive Learning for Wearable-based Human Activity Recognition

学校读者我要写书评

暂无评论

Auto-Augmentation Contrastive Learning for Wearable-based Hu...

2023 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2023

作者： Wu, Qingyu Shen, Jianfei Fan, Feiyi Gu, Yang Xu, Chenyang Chen, Yiqiang Chinese Academy of Sciences Institute of Computing Technology Beijing China University of Chinese Academy of Sciences Beijing China The Beijing Key Laboratory of Mobile Computing and Pervasive Device Beijing China Shandong Academy of Intelligent Computing Technology ShanDong China Peng Cheng Laboratory Shenzhen China Tianjin University Institute of Robotics and Autonomous Systems Tianjin China

ISBN: (纸本)9798350337488

For low-semantic sensor signals from human activity recognition (HAR), contrastive learning (CL) is essential to implement novel applications or generic models without manual annotation, which is a high-performance self-supervised learning (SSL) method. However, CL relies heavily on data augmentation for pairwise comparisons. Especially for low semantic data in the HAR area, conducting good performance augmentation strategies in pretext tasks still rely on manual attempts lacking generalizability and flexibility. To reduce the augmentation burden, we propose an end-to-end auto-augmentation contrastive learning (AutoCL) method for wearable-based HAR. AutoCL is based on a Siamese network architecture that shares the parameters of the backbone and with a generator embedded to learn auto-augmentation. AutoCL trains the generator based on the representation in the latent space to overcome the disturbances caused by noise and redundant information in raw sensor data. The architecture empirical study indicates the effectiveness of this design. Furthermore, we propose a stop-gradient design and correlation reduction strategy in AutoCL to enhance encoder representation learning. Extensive experiments based on four wide-used HAR datasets demonstrate that the proposed AutoCL method significantly improves recognition accuracy compared with other SOTA methods. © 2023 IEEE.

关键词： Contrastive Learning Human Activity Recognition Self-Supervised Learning

Machine learning based 3D object detection for navigation in unstructured environments

学校读者我要写书评

暂无评论

Machine learning based 3D object detection for navigation in...

IEEE Intelligent Vehicles Symposium Workshops (IV Workshops)

作者： Gjorgji Nikolovski Michael Reke Ingo Elsen Stefan Schiffer Mobile Autonomous Systems & Cognitive Robotics Institute FH Aachen University of Applied Sciences Aachen Germany

ISBN: (纸本)9781665479226

In this paper we investigate the use of deep neural networks for 3D object detection in uncommon, unstructured environments such as in an open-pit mine. While neural nets are frequently used for object detection in regular autonomous driving applications, more unusual driving scenarios aside street traffic pose additional challenges. For one, the collection of appropriate data sets to train the networks is an issue. For another, testing the performance of trained networks often requires tailored integration with the particular domain as well. While there exist different solutions for these problems in regular autonomous driving, there are only very few approaches that work for special domains just as well. We address both the challenges above in this work. First, we discuss two possible ways of acquiring data for training and evaluation. That is, we evaluate a semi-automated annotation of recorded LIDAR data and we examine synthetic data generation. Using these datasets we train and test different deep neural network for the task of object detection. Second, we propose a possible integration of a ROS2 detector module for an autonomous driving platform. Finally, we present the performance of three state-of-the-art deep neural networks in the domain of 3D object detection on a synthetic dataset and a smaller one containing a characteristic object from an open-pit mine.

关键词： Deep learning Training Solid modeling Three-dimensional displays Annotations Conferences Neural networks

A high-accuracy, low-budget Sensor Glove for Trajectory Model Learning 20

学校读者我要写书评

暂无评论

A high-accuracy, low-budget Sensor Glove for Trajectory Mode...

20th International Conference on Advanced robotics, ICAR 2021

作者： Denz, Robin Demirci, Rabia Cansev, M. Ege Bliek, Adna Beckerle, Philipp Rueckert, Elmar Rottmann, Nils Institute For Robotics and Cognitive Systems University of Luebeck Luebeck23562 Germany Chair of Autonomous Systems and Mechatronics Friedrich-Alexander-University Department of Electrical Engineering Erlangen-Nürnberg Germany Chair of Cyber-Physical-Systems Montanuniversität Leoben Leoben Austria

ISBN: (纸本)9781665436847

Sensor gloves are gaining importance in tracking hand and finger movements in virtual reality applications as well as in scientific research. They introduce an unrestricted way of capturing motion without the dependence on direct line of sight as for visual tracking systems. With such sensor gloves, data of complex motion tasks can be recorded and used for modeling probabilistic trajectories or teleoperation of robotic arms. While a multitude of sensor glove designs relying on different functional principles exist, these approaches require either sensitive calibration and sensor fusion methods or complex manufacturing processes. In this paper, we propose a low-budget, yet accurate sensor glove system that uses flex sensors for fast and efficient motion tracking. We evaluate the performance of our sensor glove, such as accuracy and latency, and demonstrate the functionality by recording motion data for learning probabilistic movement models. © 2021 IEEE.

关键词： Budget control

Generative Graphical Inverse Kinematics

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Limoyo, Oliver Marić, Filip Giamou, Matthew Alexson, Petra Petrović, Ivan Kelly, Jonathan Space and Terrestrial Autonomous Robotic Systems Laboratory University of Toronto Institute for Aerospace Studies Toronto Canada Laboratory for Autonomous Systems and Mobile Robotics University of Zagreb Faculty of Electrical Engineering and Computing Zagreb Croatia

Quickly and reliably finding accurate inverse kinematics (IK) solutions remains a challenging problem for many robot manipulators. Existing numerical solvers are broadly applicable but typically only produce a single solution and rely on local search techniques to minimize nonconvex objective functions. More recent learning-based approaches that approximate the entire feasible set of solutions have shown promise as a means to generate multiple fast and accurate IK results in parallel. However, existing learning-based techniques have a significant drawback: each robot of interest requires a specialized model that must be trained from scratch. To address this key shortcoming, we propose a novel distance-geometric robot representation coupled with a graph structure that allows us to leverage the sample efficiency of Euclidean equivariant functions and the generalizability of graph neural networks (GNNs). Our approach is generative graphical inverse kinematics (GGIK), the first learned IK solver able to accurately and efficiently produce a large number of diverse solutions in parallel while also displaying the ability to generalize—a single learned model can be used to produce IK solutions for a variety of different robots. When compared to several other learned IK methods, GGIK provides more accurate solutions with the same amount of data. GGIK can generalize reasonably well to robot manipulators unseen during training. Additionally, GGIK can learn a constrained distribution that encodes joint limits and scales efficiently to larger robots and a high number of sampled solutions. Finally, GGIK can be used to complement local IK solvers by providing reliable initializations for a local optimization process. Copyright © 2022, The Authors. All rights reserved.

关键词： Inverse kinematics

A Survey on Offline Reinforcement Learning: Taxonomy, Review, and Open Problems

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Prudencio, Rafael Figueiredo Maximo, Marcos R.O.A. Colombini, Esther Luna The Laboratory of Robotics and Cognitive Systems Computing Institute University of Campinas Campinas13083-852 Brazil The Autonomous Computational Systems Laboratory Computer Science Division Aeronautics Institute of Technology São José dos Campos12228-900 Brazil

With the widespread adoption of deep learning, reinforcement learning (RL) has experienced a dramatic increase in popularity, scaling to previously intractable problems, such as playing complex games from pixel observations, sustaining conversations with humans, and controlling robotic agents. However, there is still a wide range of domains inaccessible to RL due to the high cost and danger of interacting with the environment. Offline RL is a paradigm that learns exclusively from static datasets of previously collected interactions, making it feasible to extract policies from large and diverse training datasets. Effective offline RL algorithms have a much wider range of applications than online RL, being particularly appealing for real-world applications, such as education, healthcare, and robotics. In this work, we contribute with a unifying taxonomy to classify offline RL methods. Furthermore, we provide a comprehensive review of the latest algorithmic breakthroughs in the field using a unified notation as well as a review of existing benchmarks' properties and shortcomings. Additionally, we provide a figure that summarizes the performance of each method and class of methods on different dataset properties, equipping researchers with the tools to decide which type of algorithm is best suited for the problem at hand and identify which classes of algorithms look the most promising. Finally, we provide our perspective on open problems and propose future research directions for this rapidly growing field. © 2022, CC BY.

关键词： Reinforcement learning