The enormous volume of mobility data in this new era thanks to on-board devices, sensors and wireless connectivity has posed new challenges in the world of mobility big data management. However, Track&Know’s set of toolboxs, including Big Data Processing (BDP), Big Data Analytics (BDA), Complex Event Recognition (CER), Visual Analytics (VA) can enable the applications of big data to become opportunities to innovate the management and operation of fleet management systems.

BDP aims at supporting novel and scalable, solutions of high throughput addressing storage, efficient access, indexing, partitioning and load balancing for Big spatio-temporal data with reliable data collection modes and a set of big data operators. Therefore, increased number of external sources are integrated (e.g. weather, points of interest …); invalid coordinates and invalid speed calculations due to errors are reduced in the fleet monitoring system.

BDA deliver scalable trajectory data mining techniques for voluminous data and real-time techniques to incrementally capture recurring or rapidly evolving phenomena.

  • With support for computing intensive, analytic processing and machine learning techniques, BDA helps identify driving behavior excess per driver, identify patterns leading to improved fleet maintenance costs and support preventive maintenance recommendations based on tracked parameters (service downtime, etc.)
  • With Future Location Predictation, BDA helps proactively identify traffic hot spots per day and its alternative routes
  • With Trajectory Prediction, BDA helps provide recommendations for fuel consumption reduction based on the overall fleet performance optimization, provide accurate estimations of future travel distances and increase the recommendations for alternative routes based on fuel economy and road conditions.



To identify and investigate potentially dangerous driving behaviors in commercial fleet vehicles, the approach transforms big data of vehicle trajectories extracted from tracking devices to a visual analytics workflow to analyse dynamic attributes of moving vehicles before and after the event of interest. 

1st step: Selection of the events of interest with their times and locations. (e.g. harsh braking, harsh acceleration)

2nd step: Selection of the relevant attributes (e.g. speed, engine status, fuel amount …), the desired window in relation to the event time and the temporal resolution considering the sampling rate of the available data à Each event is then characterised by a vector of contextual attribute values.

3rd step: Repeated patterns are discovered by applying clustering to the vectors of all events with an appropriate similarity measure and the clustering technique. à Attribute characteristics of the clusters are presented in visual displays for comparison and semantic interpretation.

4th step: Investigation of the spatial and temporal distribution of the clusters à Identification of spatial or spatio-temporal “hot spot”.

FACT: Siming Chen and other research collaborators for their research work on  ‘’Contextualised analysis of movement events’’ that received the best paper award in EuroVA 2019 workshop held jointly with EuroVis 2019 Conference in Lisbon (Portugal) on June 3-6, 2019.