1,721,093 research outputs found
An Architecture for a Mobility Recommender System in Smart Cities
No full textTo date, there is a wide availability of academic and commercial ICT proposals to improve urban mobility. Nevertheless, in the literature, there is still a lack of suitable solutions for door-to-door routing supporting users from their origins to destinations and including the suggestion on where to park. On the other hand, in an Internet-of-Things (IoT) scenario, a lot of novel information sources could be exploited to compute more efficient mobility solutions to be proposed to the user. As an example, parking availability data could be easily collected and exploited to provide multimodal routes (i.e. routes with at least two different means of transportation) that include suggestions on where to park and how to reach the final destination. In this paper, we describe a distributed IoT architecture towards the definition of a Mobility Recommender System. In particular, we focus on a car-based multimodality, where the user always starts a trip with his/her private vehicle, but he/she can also leave the car in Park-and-Ride infrastructures and reach the destination with public transportation. This type of routing on a wider search area will result to be more costly, and thus, it will particularly benefit from a parallel computational architectural solution
Discovering Information from Spatial Big Data for ITS
No full textModern vehicles include a wide variety of sensors. With the arrival on the market of connected cars, information collected from these sensors can be potentially shared with a remote data center in real-time. This could give rise to one of the largest sensor networks in the world, harvesting gigabytes per second, which can be referred to as Spatial Big Data (SBD). The number and type of applications that could be generated on top of the knowledge extracted from the data coming by vehicles’ sensors is almost limitless, ranging from much more accurate Intelligent Transportation Systems (ITS), to better weather forecast, insight on the geographical distribution of the pollutants, and so on. On the other hand, this scenario will pose severe challenges to the practitioner in the field of Information and Communication Technologies (ICT), since the amount of SBD that could be collected by the sensor network of connected cars will by far exceed the storage and processing capacity of commonly computing and database technologies. Therefore, the key tasks of the Knowledge Discovery Process (KDP), mainly data storage and data mining, should be significantly revised in order to be able to properly handle and exploit such a big amount of spatial data. In this chapter, we discuss the current state of the art and we provide an overview of the main challenges that must be faced in the KDP domain, in order to have an infrastructure ready to deal with SBD coming from the sensors of the connected cars, and to extract new knowledge able to generate novel and exciting applications
On-street Parking Availaibilty Data in San Francisco, from Stationary Sensors and High-Mileage Probe Vehicles
Full text linkThis dataset contains records of the measured on-street parking availability in San Francisco, obtained from the public API of the SFpark project.
In 2011, the San Francisco Municipal Transportation Agency (SFMTA) started a project on smart parking, called SFpark, whose goal was the improvement of on-street parking management in San Francisco, mostly by means of demand-responsive price adjustments. One of the key points of the project was the collection of information about on-street parking availability. To this aim, about 8,000 parking spaces were equipped with specific sensors in the asphalt, periodically broadcasting availability information. The SFpark project made available a public REST API, returning the number of free parking spaces and total number of provided parking spaces per road segment, for 5,314 parking spaces on 579 road segments in the pilot area. We collected parking availability data from 2013/06/13 until 2013/07/24, by querying this API at approximately 5-minute intervals. As a result, we obtained in total about 7 million observations of parking availability on the road segments. These observations represent the first dataset we are providing.
In addition, we simulated the achievable sensing coverage of on-street parking availability that could be achieved by a fleet of taxis, if they were equipped with sensors able to detect free parking spaces, like side-scanning ultrasonic sensors, or windshield-mounted cameras [4]. In particular, by exploiting real taxi trajectories in San Francisco from the Cabspotting project, we first computed the frequencies of taxi visits for each road segment covered by the SFpark sensors. Then, we downsampled the first dataset, in order to have a parking availability information for a road segment at a given time only in presence of a transit of a taxi on that segment at that time. This step was replicated for 5 different sizes of taxi fleets, namely 100, 200, 300, 400, and 486.
Consequently, in total six datasets are available for further research in the field of on-street parking dynamics
On-street parking data in San Francisco - SFpark sensor data and simulated crowd-sensing data
No full textThis dataset contains records of the measured parking availability in San Francisco, obtained from the public API of the SFpark project. In addition, by combining the SFpark data with taxi trajectories from the Cabspotting project, we simulated parking crowd-sensing with taxis as probe vehicles
Multimodal Automotive Telematics Systems: Design and Evaluation
No full textThe definition of Automotive (or Vehicular) Telematics Systems (VTS) is a challenging task. Traditional HCI/Software Engineering approaches and methodologies cannot be totally applied, due to the number of issues inducted by the automotive domain, such as safety concerns, hardware constraints, modular architectures, etc... In this dissertation we deal with the design of both the User Interface (UI), and the software architecture design, the development and the evaluation tasks of an Automotive Telematics Systems. About the UI, our aim was to define a novel approach meant to synergistically integrate visual, auditory, and tactile information, in order to limit driver’s cognitive and visual workload inducted by VTS. As a result, we propose a new interaction device, named Handy and currently patent pending. Thanks to its specific shape and positioning, driver has always within reach all the commands needed to interact with the VTS, and could rely only on his/her tactile channel to identify the suited controls, with a significant improvement of the overall safety, as reported by experimental trials we conducted. Handy is supported by a specifically conceived Graphical User Interface (GUI), which also exhibits several distinguishing features. Indeed, it is meant to minimize the driver’s visual workload, by carefully calibrating the appropriate amount of information to display. The GUI is complemented by a Vocal User Interface (VUI), to obtain, together with Handy, an integrated, multimodal interface. The VUI is based on the command word paradigm and encompasses a new atomic dialogue paradigm, based on earcons and a help-on-demand mechanism. The defined auditory interface turns out to be smart for expert users, but also effective for unskilled people. Empirical results about its effectiveness were very satisfactory. About software architecture, the experience gained from designing and developing a prototype of VTS allowed us to define a new design pattern, useful to obtain flexible and modular architectures. It is intended as an evolution of the Model-View-Controller pattern and has been conceived taking into account aspects, such as concurrency, asynchronous communication, multimodality, and management of divergent services. Among the main advantages inducted by such a pattern there is an higher degree of modularity, enhancements in maintainability, extensibility and testability, and the decoupling between the UI of a subsystem and its underlying logic implementations. About the evaluation of VTS UI, it requires different approaches than assessment of traditional UIs, because, it is necessary not only to consider the user interaction with the interface but also to understand the effects of this interaction on driver-vehicle performances. Thus, it requires the user is focused on the primary driving task, and concurrently interact with the system. To this aim, we propose a new a framework specifically suited for indoor evaluation of VTSs UI. Such a framework allowed us to run empirical analysis on a sample of end-users, to assess the effectiveness of the proposed multimodal UI, in terms of driver-vehicle performances. The results gathered on a sample of eight subjects were very encouraging, since they report lower driver’s workload, than other concurrent automotive solutions
On the Impact of Location-related Terms in Neural Embeddings for Content Similarity Measures in Cultural Heritage Recommender Systems
No full textAnalysing text to detect semantic similarities is a recent breakthrough of Natural Language Processing that brought many novel applications in different fields. A domain which could greatly benefit of this innovation is the one regarding Location-based and/or Touristic Recommender Systems, where the user receives suggestions based on his/her past liked items. In this work, we consider the use of neural embeddings weighted using Smooth-Inverse Frequency (SIF) to detect semantic similarities in textual descriptions found in a large graph database covering Italian cultural Points of Interests (POIs). Of all detected similar pairs on a national scale, 19% are composed by POIs that do not belong to the same ontological category, highlighting the potential neural embeddings have to match POIs beyond the categories they have been assigned to. However, since text descriptions also contain references to the places where POIs are found, similarities can be detected among POIs sharing the same location, especially in the case of low-frequency geographical terms. While this may be desirable, in some cases, it may harm location-aware applications, as POIs positions are already known. By comparing city names occurrence probabilities both in the full text corpus and in location-constrained sub-corpora, we observed probability shifts, on average, of 232%. This suggests that, for the specific case of location-aware services, SIF-weighted neural embeddings should use location-constrained sub-corpora for term occurrence probability computation in order to efficiently remove uninteresting information
11th International Symposium on Web and Wireless Geographical Information Systems (W2GIS 2012).
No full text
- …
