Detecting Separation in Robotic and Sensor Networks
Aerospace Engineering, University of Florida, Gainesville, FL.;Engineering, Texas A\&M University, College Station, TX.;Dept. of Aerospace Engineering, Texas A\&M University, College Station, TX.
Chenda Liao, Harshavardhan Chenji, Prabir Barooah, Radu Stoleru, Tamás Kalmár-Nagy
In this paper we consider the problem of monitoring detecting separation of agents from a base station in robotic and sensor networks. Such separation can be caused by mobility and/or failure of the agents. While separation/cut detection may be performed by passing messages between a node and the base in static networks, such a solution is impractical for networks with high mobility, since routes are constantly changing. We propose a distributed algorithm to detect separation from the base station. The algorithm consists of an averaging scheme in which every node updates a scalar state by communicating with its current neighbors. We prove that if a node is permanently disconnected from the base station, its state converges to $0$. If a node is connected to the base station in an average sense, even if not connected in any instant, then we show that the expected value of its state converges to a positive number. Therefore, a node can detect if it has been separated from the base station by monitoring its state. The effectiveness of the proposed algorithm is demonstrated through simulations, a real system implementation and experiments involving both static as well as mobile networks.
AHA: A Vision-Language-Model for Detecting and Reasoning Over Failures in Robotic Manipulation
NVIDIA(英伟达);University of Washington(华盛顿大学);Universidad Católica San Pablo(圣巴勃罗天主教大学);MIT(麻省理工学院);Nanyang Technological University(南洋理工大学);Allen Institute for Artificial Intelligence(艾伦人工智能研究所)
Trojan Attacks on Neural Network Controllers for Robotic Systems
Concordia University(康科德大学);Concordia Institute for Information Systems Engineering(康科德信息系统工程研究所);Fonds de recherche du Québec – Nature et Technologies(魁北克自然与技术研究基金);National Cybersecurity Consortium(国家网络安全联盟)
Real-Time, Highly Accurate Robotic Grasp Detection using Fully Convolutional Neural Networks with High-Resolution Images
机器人抓取检测新物体是一项具有挑战性的任务,但在过去的几年里,基于深度学习的方法已经取得了显着的性能改进,使用RGB-D数据的准确率高达96.1%。在本文中,我们提出了完全卷积神经网络(FCNN)为基础的机器人抓取检测方法。我们的方法还实现了最先进的检测精度(高达96.6%),并在Cornell数据集上对高分辨率图像(每360 x 360图像6- 20 ms)进行了最先进的实时的【形】计算时间。由于FCNN,我们提出的方法可以应用于任何大小的图像,用于检测多物体上的多抓取。提出的方法进行了评估,使用4轴机械臂与小型平行夹持器和RGB-D相机抓取具有挑战性的小,新的对象。通过我们提出的基于学习的全自动方法进行精确的视觉机器人坐标校准,我们提出的方法取得了90%的成功率。
Developing Simulation Models for Soft Robotic Grippers in Webots
Faculty of Science and Engineering, University of Groningen(格罗宁根大学科学与工程学院);School of Electrical Engineering and Informatics, Bandung Institute of Technology(班达理工大学电气工程与信息学院)
Unveiling the Potential of iMarkers: Invisible Fiducial Markers for Advanced Robotics
Automation and Robotics Research Group, Interdisciplinary Centre for Security, Reliability, and Trust (SnT), University of Luxembourg(自动化与机器人研究组,安全、可靠性与信任跨学科中心(SnT),卢森堡大学);Faculty of Science, Technology, and Medicine, University of Luxembourg(科学、技术与医学学院,卢森堡大学);Department of Physics & Materials Science, University of Luxembourg(物理与材料科学系,卢森堡大学);Institute for Advanced Studies, University of Luxembourg(先进研究学院,卢森堡大学)