The recent development in autonomous driving involves high-level computer vision and detailed road scene understanding.Today,most autonomous vehicles employ expensive high quality sensor-set such as light detection an...The recent development in autonomous driving involves high-level computer vision and detailed road scene understanding.Today,most autonomous vehicles employ expensive high quality sensor-set such as light detection and ranging(LIDAR)and HD maps with high level annotations.In this paper,we propose a scalable and affordable data collection and annotation framework image-to-map annotation proximity(I2MAP),for affordance learning in autonomous driving applications.We provide a new driving dataset using our proposed framework for driving scene affordance learning by calibrating the data samples with available tags from online database such as open street map(OSM).Our benchmark consists of 40000 images with more than40 affordance labels under various day time and weather even with very challenging heavy snow.We implemented sample advanced driver-assistance systems(ADAS)functions by training our data with neural networks(NN)and cross-validate the results on benchmarks like KITTI and BDD100K,which indicate the effectiveness of our framework and training models.展开更多
Fast data synchronization in wireless ad hoc networks is a challenging and critical problem.It is fundamental for efficient information fusion,control and decision in distributed systems.Previously,distributed data sy...Fast data synchronization in wireless ad hoc networks is a challenging and critical problem.It is fundamental for efficient information fusion,control and decision in distributed systems.Previously,distributed data synchronization was mainly studied in the latency-tolerant distributed databases,or assuming the general model of wireless ad hoc networks.In this paper,we propose a pair of linear network coding(NC)and all-to-all broadcast based fast data synchronization algorithms for wireless ad hoc networks whose topology is under operator’s control.We consider both data block selection and transmitting node selection for exploiting the benefits of NC.Instead of using the store-and-forward protocol as in the conventional uncoded approach,a compute-and-forward protocol is used in our scheme,which improves the transmission efficiency.The performance of the proposed algorithms is studied under different values of network size,network connection degree,and per-hop packet error rate.Simulation results demonstrate that our algorithms significantly reduce the times slots used for data synchronization compared with the baseline that does not use NC.展开更多
In recent years, cloud sync(hronization) services such as GoogleDrive and Dropbox have provided Internet users with convenient and reliable data storing/sharing functionality. The cloud synchronization mechanism (i...In recent years, cloud sync(hronization) services such as GoogleDrive and Dropbox have provided Internet users with convenient and reliable data storing/sharing functionality. The cloud synchronization mechanism (in particular, how to deliver the user-side data updates to the cloud) plays a critical role in cloud sync services because it greatly affects the cloud operation cost (in terms of sync traffic) and user experience (in terms of sync delay). By comprehensively measuring tens of popular cloud sync services, we find that their cloud sync mechanisms differ greatly in sync performance and design granularity. Quite surprisingly, some very popular services (like GoogleDrive and 115 SyncDisk) utilize a quite coarse-grained cloud sync mechanism that may lead to severe traffic overuse. For example, updating 1-MB data may sometimes result in 260-MB sync traffic. In this paper, we conduct a comparative study of various cloud sync mechanisms by analyzing their respective pros and cons under different situations, unravel the pathological processes for their traffic overuse problems, and finally provide insights/solutions for better choosing/designing a cloud sync service.展开更多
The structure of a power energy system is becoming more distributed than before.It becomes challenging to manage such a system in a centralized way,because a central authority may not exist or may not be trusted by al...The structure of a power energy system is becoming more distributed than before.It becomes challenging to manage such a system in a centralized way,because a central authority may not exist or may not be trusted by all parties.Blockchain is a promising tool to address this challenge,by enabling trusted collaboration in the absence of a trusted central authority.Its use in the energy sector has been pioneered by several pilot projects.However,to date the energy sector has not seen large-scale deployment of blockchain,partly because the founders of those pilot projects,the public,and utilities have not reached consensus on the values and limitations of blockchain in energy.This perspective aims to bridge this gap.First,the philosophy and unique values of blockchain are discussed.Second,some promising blockchain-based applications in energy systems are presented.Third,some common misunderstandings of blockchain in energy are discussed.Last,some frequently-asked questions from utilities are discussed.Hopefully this perspective can help advance large-scale deployment of blockchain in energy systems.展开更多
文摘The recent development in autonomous driving involves high-level computer vision and detailed road scene understanding.Today,most autonomous vehicles employ expensive high quality sensor-set such as light detection and ranging(LIDAR)and HD maps with high level annotations.In this paper,we propose a scalable and affordable data collection and annotation framework image-to-map annotation proximity(I2MAP),for affordance learning in autonomous driving applications.We provide a new driving dataset using our proposed framework for driving scene affordance learning by calibrating the data samples with available tags from online database such as open street map(OSM).Our benchmark consists of 40000 images with more than40 affordance labels under various day time and weather even with very challenging heavy snow.We implemented sample advanced driver-assistance systems(ADAS)functions by training our data with neural networks(NN)and cross-validate the results on benchmarks like KITTI and BDD100K,which indicate the effectiveness of our framework and training models.
基金This work is financially supported by Beijing Municipal Natural Science Foundation(No.L202012)the Open Research Project of the State Key Laboratory of Media Convergence and Communication,Communication University of China(No.SKLMCC2020KF008)the Fundamental Research Funds for the Central Universities(No.2020RC05).
文摘Fast data synchronization in wireless ad hoc networks is a challenging and critical problem.It is fundamental for efficient information fusion,control and decision in distributed systems.Previously,distributed data synchronization was mainly studied in the latency-tolerant distributed databases,or assuming the general model of wireless ad hoc networks.In this paper,we propose a pair of linear network coding(NC)and all-to-all broadcast based fast data synchronization algorithms for wireless ad hoc networks whose topology is under operator’s control.We consider both data block selection and transmitting node selection for exploiting the benefits of NC.Instead of using the store-and-forward protocol as in the conventional uncoded approach,a compute-and-forward protocol is used in our scheme,which improves the transmission efficiency.The performance of the proposed algorithms is studied under different values of network size,network connection degree,and per-hop packet error rate.Simulation results demonstrate that our algorithms significantly reduce the times slots used for data synchronization compared with the baseline that does not use NC.
基金supported in part by the National Natural Science Foundation of China (No. 61073015)the National Key Basic Research and Development (973) Program of China (No. 2011CB302305)National Key Projects of Science and Technology of China (No.2010ZX03004-001-03)
文摘In recent years, cloud sync(hronization) services such as GoogleDrive and Dropbox have provided Internet users with convenient and reliable data storing/sharing functionality. The cloud synchronization mechanism (in particular, how to deliver the user-side data updates to the cloud) plays a critical role in cloud sync services because it greatly affects the cloud operation cost (in terms of sync traffic) and user experience (in terms of sync delay). By comprehensively measuring tens of popular cloud sync services, we find that their cloud sync mechanisms differ greatly in sync performance and design granularity. Quite surprisingly, some very popular services (like GoogleDrive and 115 SyncDisk) utilize a quite coarse-grained cloud sync mechanism that may lead to severe traffic overuse. For example, updating 1-MB data may sometimes result in 260-MB sync traffic. In this paper, we conduct a comparative study of various cloud sync mechanisms by analyzing their respective pros and cons under different situations, unravel the pathological processes for their traffic overuse problems, and finally provide insights/solutions for better choosing/designing a cloud sync service.
基金supported by the National Key R&D Program of China(No.2021YFC2100100)the National Natural Science Foundation of China(No.21901157)the Shanghai Science and Technology Project of China(No.21JC1403400).
文摘The structure of a power energy system is becoming more distributed than before.It becomes challenging to manage such a system in a centralized way,because a central authority may not exist or may not be trusted by all parties.Blockchain is a promising tool to address this challenge,by enabling trusted collaboration in the absence of a trusted central authority.Its use in the energy sector has been pioneered by several pilot projects.However,to date the energy sector has not seen large-scale deployment of blockchain,partly because the founders of those pilot projects,the public,and utilities have not reached consensus on the values and limitations of blockchain in energy.This perspective aims to bridge this gap.First,the philosophy and unique values of blockchain are discussed.Second,some promising blockchain-based applications in energy systems are presented.Third,some common misunderstandings of blockchain in energy are discussed.Last,some frequently-asked questions from utilities are discussed.Hopefully this perspective can help advance large-scale deployment of blockchain in energy systems.