Search

This work describes the early stage of an interactive and accelerated AI-driven framework for Practical Driving Courses and Driving Licence Exams. The core of the project is an innovative multi-parameter AI-assisted telemetry system able to compute test scores and outcome, useful for human-neutral auditability of Driving Licence Exams. The distributed Artificial Intelligence (AI) system available at the Track Testbed will be able to perform driving behaviour classifications and will suggest specific improvements based on the analysis of vehicle trajectories acquired during the driving test. Finally, the project will target the creation of a large dataset for driving test classification of key performance parameters. The system is envisioned to have a relevant impact on all the certification, driving licence operators and regulator entities.

ith the rapid increasing availability of information and popularization of mobility devices, trajectories have become more complex in their form. Trajectory data is now high dimensional, and often associated with heterogeneous sources of semantic data, that are called Multiple Aspect Trajectories. The high dimensionality and heterogeneity of these data makes classification a very challenging task both in term of accuracy and in terms of efficiency. The present demo offers a tool, called AUTOMATISE, to support the user in the classification task of multiple aspect trajectories, specifically for extracting and visualizing the movelets, the parts of the trajectory that better discriminate a class. The AUTOMATISE integrates into a unique platform the fragmented approaches available in the literature for multiple aspects trajectories and, in general, for multidimensional sequence classification into a unique web-based and python library system. We illustrate the architecture and the use of the tool for offering both movelets visualization and a complete configuration of classification experimental settings.

With the ever burgeoning traffic and increasing unwarranted lane changes, safety of commuters and driving practices have been compromised. Most of the accidents that take place are due to unforeseen traffic and road conditions or misjudgement by the drivers. To aid the decision making of drivers to maneuver the vehicle safely, this paper proposes ELaNES, a novel system focusing on the lane changing behaviour of a vehicle. It posits a method to predict the lane change and state the characteristics associated with it. Along with this, a degree of safe lane change is calculated by taking into consideration the neighbouring vehicles’ dynamics and the driver’s intentions. The results obtained are promising, accurate and can help make decisions in the real world scenario, along with giving deep insights into human driving practices.

The application of eye-tracking technology in construction safety management hazards is studied to accurately understand the personnel’s cognition of construction hazards. Based on OSHA’s construction safety training materials, this study aims to apply the eye-tracking technology to construction safety into virtual reality (VR) scene. The eye-tracking technology is combined with focus time, trajectory analysis, and other indicators. The tester validated the technology on the construction hazard. Finally, the research results are compared to OSHA construction hazard identification training materials, and feasible development suggestions and directions in the future are suggested.

This paper introduces novel minimum-deviation control methods to improve step-load transient responses in non-inverting (four-switch) buck-boost dc-dc converters. The approach utilizes all possible switching states to achieve sub-stantial transient response improvements compared to buck-only or boost-only minimum-deviation or time-optimal control (TOC) methods. Using state-plane trajectory analysis, analytical solutions with necessary conditions are developed for step-up and step-down load transient scenarios in both buck and boost modes of operation. An experimental low-voltage prototype intended for mobile applications is used for experimental verification. High effectiveness of the proposed solution is verified by circuit simulations and experiments.

This paper deals with ways of considering dynamic stability with possibilities of its assessment by means of simulation computing, which includes parameters simulating the actual operation of the power system with all its elements.

With the continuous progress of science and technology, people's production and lifestyle which has lasted for thousands of years are also changing quietly. More and more things, people, data, and the Internet are connected together, which also makes the power of big data calculation and mining increase exponentially. The navigation data of Zhejiang coastal fishermen are recorded by satellite. Using big data calculation can combine the ocean climate information, fishing vessel location information, and the data formed by fishing grounds into a big fishery data platform. After many captains use the data provided by this platform for fishing guidance, they have achieved a very good catch. They use scientific and technological innovation to change fishing methods so that fishermen can no longer rely on heaven for food. The risk is difficult to predict. The research in this paper is to find the common characteristics of fishing grounds by analyzing the navigation data of fishermen in the Zhejiang coastal area recorded by satellite and at the same time, expand the search range to estimate the new fishing ground location. This work can be used to mine data by using convolutional neural network.

For a long time, traffic surveillance cameras have been of great help in dealing with traffic incidents, monitoring, and analyzing driver behavior, and urban traffic planning. However, traffic monitoring video data are vast and so often have a high computational cost in post analysis and statistics, and the responsiveness is also an issue. In order to more comprehensively obtain vehicle behavior information at intersections, more and more surveillance cameras are being replaced by fisheye cameras, which makes traditional vehicle identification and tracking more difficult. Therefore, this study proposes a system architecture designed for the management of fisheye video data at intersections in scenarios of police traffic monitoring. Automatically identifying and tracking vehicles and processing their trajectories into subsequent semi-structured data that can be easily searched and counted, increases the possibilities of data usage.

This paper extends the Nyquist trajectory analysis for Fractional Proportional Resonant controllers applied to Current-Controlled Voltage-Source-Inverters in the continuous-time domain. The objective is to analyse the stability and robustness of the system under the use of this controller-type. Applying frequency specifications new analytical expressions are obtained and used to tune the controller.

Data exploration is an essential task for gaining an understanding of the potential and limitations of novel datasets. This paper discusses the challenges related to exploring large Automatic Identification System (AIS) datasets. We address these challenges using trajectory-based analysis approaches implemented in distributed computing environments using Spark and GeoMesa. This approach enables the exploration of datasets that are too big to handle within conventional spatial database systems. We demonstrate our approach using a case study of 4 billion AIS records.

Security awareness is becoming more and more critical in modern society. In this paper, we propose an infrared pedestrian tracking and trajectory analysis system. The system is able to monitor large area like factories and communities, and provides monitoring and suspicious alarm functions. An attention based object detector is structured to integrate features from different level. Deep-sort style tracker is used by following the principle of tracking-by-detection. Based on the tracking results, the system provides the function of suspect alarm by analyzing the trajectories. The performance of pedestrian tracking and suspect alarm satisfies the industrial needs in the experiment.

In this paper, we propose a ball position-measurement method to evaluate the performance of a baseball pitching trajectory analysis system. To evaluate the performance of the trajectory analysis system, a method of accurately measuring the actual ball position is needed. Existing methods cannot perform a highly accurate, continuous performance evaluation of the pitching trajectory analysis system. The proposed method accurately measures ball position using measuring equipment by passing the pitch through a structure in which two cameras are installed. The results demonstrated that the error was smaller than the radius of the ball at 40 points in the structure, and the system can be used for performance evaluation of the pitching trajectory analysis system.

Despite the advancement of Internet-of-Things (IoT) and social network, one of the main challenges in SIoT domain is intelligent service discovery and composition. This paper presents an SIoT architecture with personalized recommendation in order to deliver better service discovery. Our proposed approach has outperformed other methods during the experiments.

The increasing availability of GPS-embedded devices has introduced a new dimension in digital market especially location-based services. In practice, the location data is used to understand and predict consumer mobility behavior and trend for various purposes. In this paper, we propose two methodologies to first identify suspicious location from consumer location data and to infer location at both individual device and device to device level based on systematic solution. Using stay-point clustering and suspicious patterns we identified from extensive analysis, 20-30% of records with location were observed to be suspicious. After removing inaccurate location data, we have employed scalable heuristic approach to backfill records with location even for devices that originally had no available location. Our model showed the accuracy within 50 meters at 95th percentile across different countries, including Japan, Indonesia, India, and the United States with 10-15% increase in the number of records with location and 5-10% increase in new number of devices with location.

A deep-learning-based Loitering Detection System (LDS) with re-identification (ReID) capability over a multicamera network is proposed. The proposed LDS is mainly comprising of object detection and tracking, loitering detection, feature extraction, camera switching, and re-identification of the loiterer. The person is detected using You Only Look Once (YOLOv3) and tracked using Simple Online Real-time Tracking with a deep association matrix (DeepSORT). From the trajectory analysis, once the time and displacements thresholds are satisfied, the person is treated as a loiterer. When the loiterer moves one camera to another, then the algorithm is switched to the appropriate camera feed as per the proposed camera switching algorithm to minimize the computational cost. Subsequently, the loiterer is reidentified in the switched camera feed by comparing the features of the loiterer extracted by the MobileNets with that of the other detected persons based on the triplet loss criteria. The proposed system provides an enhanced accuracy of 96 % on average fps of 33 (without ReID) and 81.5 % at average fps of 30 (with ReID).

Robots that share daily environments with us are required to behave in a socially acceptable manner. There are two important approaches to this purpose: 1) robots model human behavior, understand it properly and behave appropriately 2) robots present their understanding and future behavior to surrounding people. In this paper, considering people present various cues to other people around them using gaze and face direction, we focus on the latter approach and propose a robot that presents cues to an opposing pedestrian by turning face. Another problem with the conventional research is that the evaluation of the pedestrian’s ease of passing with the robot depends only on the subjective impression, so it was difficult to design the robot’s behavior based on the temporal change of the ease of walking. In this paper, we evaluate the fluctuation of the pedestrian’s moving velocity vector as an index of the ease of walking and analyze the temporal change. We have conducted preliminary experiments in which 12 subjects passed by the robot and compared the three types of presentation methods using the face. By presenting information using a face, we confirmed that the subjects tended to have better impressions of walking based on subjective evaluation and that the walking was relatively easy to walk for several seconds while approaching the robot based on analyzing the fluctuation of the moving speed vector.

The phenomenon of eSports has been increasing over the years, including the interest in online video games, by players, coaches and analysts. One of the most popular genres of these games is the Multiplayer Online Battle Arena (MOBA) and games like League of Legend (LoL) have increased its player base and event viewership. Similarly to other sports, players and coaches analyze the games and all the games events, such as players’ movements and spatial, temporal events in which they participate, to understand how they play and to define new strategies to help them progress. This paper presents the VisuaLeague II tool that helps players and coaches to analyze LoL game data using animated maps and other visualizations; and presents a comparative study between VisuaLeague II and other two tools (LoL replay system and OP.GG) to understand if the proposed visualization techniques are adequate for analyzing player’s performance. The results suggest that players are capable of visualizing game-related data through the use of animated maps, to solve common analytic tasks and the comparison suggests that participants prefer to visualize spatio-temporal data dynamically, i.e., using animated maps or the replay system.

This paper addresses the trajectory analysis, mission design, and control law for multiple microsatellites to cooperatively circumnavigate a host spacecraft. The cooperative circumnavigation (CCN) problem is first defined to drive a group of microsatellites to a predefined planar ellipse concerning a host spacecraft while maintaining a geometric formation configuration. Nest, a CCN control law is proposed, where the artificial potential functions and the affine Laplacian matrix are applied to fulfill the requirements of a CCN mission. Finally, simulation results are given to show the feasibility of this approach.

A baseball pitching trajectory analysis system is used to train pitchers and provide baseball content for viewers. Previous systems needed a system manager because the trajectory was detected manually, thus system automation is essential for popularizing baseball pitching trajectory analysis systems, and detecting the pitching start and end times comprise the key point toward achieving this. In this paper, we define the pitching start and end times according to the situations in baseball games and accurately detect these points in time based on a physical model. Experiments on image data captured at a professional baseball game indicate that the proposed method could detect the start and end times of a pitch as accurately as a human can.

The wide application of trajectory analysis research greatly promotes the trajectory dataset publishing which in the meantime carries out tremendous privacy leakage problems, especially those come with long-term pseudonyms. Mix-zone is a typical privacy protection approach in road network that enables a group of users switch to new pseudonyms in certain area. However, to ensure unlinkability between all incoming and all outgoing sub-trajectories, strict restrictions are defined which limits the number of qualified mix-zones. Additionally, the privacy of sub-trajectories is not further considered. In this paper, we propose a flexible trajectory partition scheme called Mix-STC which can find out more candidate mix-zones by means of three-dimensional spatiotemporal cubes calculation. Mix-STC provides privacy protection by limiting the length of sub-trajectories and perturbing the precise location of its end points. Experiments show that Mix-STC performs much better than original data in two different types of de-anonymization attacks. Our performance experiments demonstrate that the datasets processed by Mix-STC can still support typical trajectory applications well.

This paper is about the ball trajectory analysis of a multimode cricket bowling machine. The bowling machine is used to provide training for batsmen by emulating bowler's behaviour. This, in turn, reduces the risk of injuries experienced by bowlers. The experiments were divided into three groups namely ball velocity, bowling distance, and ball spin rate. For ball velocity under three machine speed settings, the measured data between dry and wet ball showed a difference in magnitude but with similar consistency. The optimal wheel pressure to produce the best dry ball velocity among those tested was 6.25 psi. The effect of speed ratio on the ball velocity was also observed from the measured velocity data where the ball velocity increase, as wheel's rotational velocity increases, peaked at 2000 rpm before decreasing. From the comparison between measured bowling length and line to that of calculations, the suitable drag coefficient value was 0.4 for estimating the ball trajectory. From the analysis, ball wetness, wheel pressure, and speed ratio were determined to be important in evaluating the performance of the cricket bowling machine.

This work involves development and implementation of an algorithm for a search of points of interest and also the identification of various patterns in the trajectories of moving objects on the basis of the association rule mining algorithms.

Biological networks are dynamic and modular. Identifying dynamic functional modules is key to elucidating biological insight and disease mechanism. In recent years, while most researchers have focused on detecting functional modules from static protein-protein interaction (PPI) networks where the networks are treated as static graphs derived from aggregated data across all available experiments or from a single snapshot at a particular time, temporal nature of context-specific transcriptomic and proteomic data has been recognized by researchers. Meanwhile, the analysis of dynamic networks has been a hot topic in data mining and social networks. Dynamic networks are structures with objects and links between the objects that vary in time. Temporary information in dynamic networks can be used to reveal many important phenomena such as bursts of activities in social networks and evolution of functional modules in protein interaction networks. In this talk, I will address several critical challenges to construct robust, dynamic gene interaction networks, and present our computational approaches to identify disease-relevant functional modules and to track the progression patterns of modules in dynamic biological networks. Significant modules which are correlated to phenotypes of interest can be identified, for example, those functional modules which form and progress across different stages of a cancer. Through identifying these functional modules in the progression process, we are able to detect the critical groups of proteins that are responsible for the transition of different cancer stages. Our approaches can also discover how the strength of each detected modules changes over the entire observation period. I will also demonstrate the application of our approach in a variety of biomedical applications.

Badminton is one of the most popular racket sports in Malaysia. In the past, badminton player movement analyses such as smashing, service, racket swing and shuttle trajectory etc. were mainly carried out qualitatively with substantial human involvement. In recent time, there is increasing assimilation of computer technology in badminton movement analysis as revealed by our review of the state-of-the-art researches. This paper aims to lay a solid foundation for the exploration of further research on badminton movement analysis. Three key areas reviewed were badminton smashing analysis, badminton service recognition, and badminton swing and shuttle trajectory analysis.

The efficiency optimization control (EOC) in steady state optimizes the flux to improve the efficiency of linear induction machine (LIM). However, it is inefficient in dynamic operations, in which the aforementioned flux solution no longer satisfies the change of thrust, thus the efficiency is deteriorated. In this paper, LIM losses including both copper and core losses during thrust transient are analyzed, and a dynamic loss model of LIM is established. A numerical solution to estimate the optimal flux trajectory is derived, and a practical method based on experimental prototype is put forward to minimize the losses as well as to meet the desired thrust response.