Abstract

This chapter explores some hypothetical computer vision pipeline designs to understand HW/SW design alternatives and optimizations. Instead of looking at isolated computer vision algorithms, this chapter ties together many concepts into complete vision pipelines. Vision pipelines are sketched out for a few example applications to illustrate the use of different methods. Example applications include object recognition using shape and color for automobiles, face detection and emotion detection using local features, image classification using global features, and augmented reality. The examples have been chosen to illustrate the use of different families of feature description metrics within the Vision Metrics Taxonomy presented in Chap. 5. Alternative optimizations at each stage of the vision pipeline are explored. For example, we consider which vision algorithms run better on a CPU versus a GPU, and discuss how data transfer time between compute units and memory affects performance. Document type: Part of book or chapter of book

Abstract

Studies have estimated that irregular operations (flights affected by a disruption) can cost between 2% and 3% of the airline annual revenue and that a better recovery process could result in cost reductions of at least 20%. Even for small airlines this can represent millions of Euros. In this paper we propose a multi-agent system (MAS) whose members represent the roles, functionalities and competences existing in a typical Airline Operations Control Centre (AOCC), the airline entity responsible for managing the impact of irregular events on planned operations. This multiagent based system produces intelligent solutions in the sense that its outcomes are the result of an autonomous reaction and adaption to changes in the environment, solving partial problems simultaneously. We tested our MAS using real data from TAP Portuguese airline company and experimentally compared our system with solutions found by the human operators on TAP Portugal AOCC. A comparison was also made with a more traditional sequential approach that is the typical method followed by AOCCs when solving disruptions. Results from those comparisons show that it is possible to reduce costs and have a better integrated solution with the proposed system.

Abstract

It is well known that airport surface traffic congestion at major airports is responsible for increased taxi-out times, fuel burn and excess emissions and there is potential to mitigate these negative consequences through optimizing airport surface traffic operations. Due to a highly congested voice communication channel between pilots and air traffic controllers and a data communication channel that is used only for limited functions, one of the most viable near-term strategies for improvement of the surface traffic is issuing a push back advisory to each departing aircraft.This dissertation focuses on the optimization of a push back time window for each departing aircraft. The optimization takes into account both spatial and temporal uncertainties of ramp area aircraft trajectories. The uncertainties are described by a stochastic kinematic model of aircraft trajectories, which is used to infer distributions of combinations of push back times that lead to conflict among trajectories from different gates. The model is validated and the distributions are included in the push back time window optimization. Under the assumption of a fixed taxiway spot schedule, the computed push back time windows can be integrated with a higher level taxiway scheduler to optimize the flow of traffic from the gate to the departure runway queue. To enable real-time decision making the computational time of the push back time window optimization is critical and is analyzed throughout.

Abstract

Traffic congestion is increasing in almost all large cities, leading to a number of negative effects such as pollution and delays. However, building new roads is not a feasible solution. Instead, the use of the existing road network has to be optimized, together with a shift towards more sustainable transport modes. In order to achieve this there are several challenges that needs to be addressed. One challenge is the ability to provide accurate information about the current and future traffic state. This information is an essential input to the traffic management center and can be used to influence the choices made by the travelers. Accurate information about the traffic state on highways, where the potential to manage and control the traffic in general is very high, would be of great significance for the traffic managers. It would help the traffic managers to take action before the system reaches congestion and limit the effects of it. At the same time, the collection of traffic data is slowly shifting from fixed sensors to more probe based data collection. This requires an adaptation and further development of the traditional traffic models in order for them to handle and take advantage of the characteristics of all types of data, not just data from the traditionally used fixed sensors. The objective of this thesis is to contribute to the development and implementation of a model for estimation and prediction of the current and future traffic state and to facilitate an adaptation of the model to the conditions of the highway in Stockholm. The model used is a version of the Cell Transmission Model (CTM-v) where the velocity is used as the state variable. Thus, together with an Ensemble Kalman Filter (EnKF) it can be used to fuse different types of point speed measurements. The model is developed to run in real-time for a large network. Furthermore, a two-stage process used to calibrate the model is implemented. The results from the calibration and validation show that once the model is calibrated, the estimated travel times corresponds well with the ground truth travel times collected from Bluetooth sensors. In order to produce accurate short-term predictions for various networks and conditions it is vital to combine different methods. We have implemented and evaluated a hybrid prediction approach that assimilates parametric and non-parametric short-term traffic state prediction. To predict mainline sensor data we use a neural network, while the CTM-v is ran forward in time in order to predict future traffic states. The results show that both the hybrid approach and the CTM-v prediction without the additional predicted mainline sensor data is superior to a naïve prediction method for longer prediction horizons.

Abstract

air traffic is forecasted to increase, air traffic control software subsequently needs to be more sophisticated. To efficiently push development forward, testing is important in order to determine usability. The tests need to be adapted to fit a particular purpose and carried out with methods that preserve the validity of the results. This thesis describes an implementation project carried out at the EUROCONTROL Experimental Centre, Bretigny-sur-Orge, France. The purpose of the project is to create an application that enables a user to create datasets of air traffic to be used for these tests. The application allows for manual work or bulk imports from external data sources. Furthermore it compiles scenarios as output datasets intended for prototype air traffic control software developed at Linköping University. The application design rationale and development process is described. Some time is spent on demonstrating the flexibility of the application and how its usage fits in a bigger picture.

Abstract

There are many map providers on the market today. Anyone who wishes to use a licensed map-service in an application has to pay a license fee. This fee can become a big expense and affect the price that the end costumer has to pay.This thesis has investigated in how to set up and improve an open source geocoding service for it to measure against a licensed map-service. Geocoding is the technique of having an input address and returning a position which consists of a latitude and a longitude coordinate. The investigation has been done by implementing an open source solution as a proof of concept with the goal to answer the question to which extent is it feasible to develop an open source geocoding service to be as fast accurate and complete as a licensed map service.The open source solution has been developed in collaboration with TaxiCaller Nordic AB. In the implementation the Pelias project has been used as a geocoder together with map data from the datasets of OpenStreetMap and Who’s On First and Elasticsearch as the search engine. The work is based on functional, data and performance requirements set by TaxiCaller.The evaluation has shown that most of the requirements set for this work are achieved with the implemented open source geocoding service solution. Examples of these requirements are correctness of the search results and that the address, street, venue or intersection in the search results should be fully specified.The functional requirement to convert an intersection to coordinates is not achieved when the intersection can not be uniquely identified. The performance requirement to search for a venue is not. Also, the data requirement that the postal code in the search results should be fully specified is not achieved.Sometimes but not always a licensed map-service can provide better data. 目前市场上有许多地图提供商。任何希望在应用程序中使用许可地图服务的人都必须支付许可费。这笔费用可能会成为一笔巨大的费用，并影响最终客户必须支付的价格。 本文研究了如何建立和改进开源地理编码服务，以便对许可的地图服务进行测量。地理编码是具有输入地址并返回由纬度和经度坐标组成的位置的技术。调查是通过实施开源解决方案作为概念证明来完成的，其目的是回答开发开源地理编码服务的可行程度，以及与许可地图服务一样快速准确和完整的问题。 开源解决方案是与TaxiCaller Nordic AB合作开发的。在实现中，Pelias项目已被用作地理编码器以及来自OpenStreetMap和Who's On First以及Elasticsearch数据集的地图数据作为搜索引擎。这项工作基于TaxiCaller设定的功能，数据和性能要求。 评估表明，通过实施的开源地理编码服务解决方案，可以实现为此项工作设置的大部分要求。这些要求的示例是搜索结果的正确性，并且应完全指定搜索结果中的地址，街道，地点或交叉点。 当无法唯一地识别交叉点时，不能实现将交叉点转换为坐标的功能要求。搜索场地的性能要求不是。此外，未实现搜索结果中的邮政编码应完全指定的数据要求。 有时但并非总是获得许可的地图服务可以提供更好的数据。

Abstract

Using multiple datacenters allows for higher availability, load balancing and reduced latency to customers of cloud services. To distribute multiple copies of data, cloud providers depend on inter-datacenter WANs that ought to be used efficiently considering their limited capacity and the ever-increasing data demands. In this paper, we focus on applications that transfer objects from one datacenter to several datacenters over dedicated inter-datacenter networks. We present DCCast, a centralized Point to Multi-Point (P2MP) algorithm that uses forwarding trees to efficiently deliver an object from a source datacenter to required destination datacenters. With low computational overhead, DCCast selects forwarding trees that minimize bandwidth usage and balance load across all links. With simulation experiments on Google's GScale network, we show that DCCast can reduce total bandwidth usage and tail Transfer Completion Times (TCT) by up to $50\\%$ compared to delivering the same objects via independent point-to-point (P2P) transfers. Comment: 9th USENIX Workshop on Hot Topics in Cloud Computing, https://www.usenix.org/conference/hotcloud17/program/presentation/noormohammadpour Document type: Conference object