Previous Abstract Return to Session C5 Next Abstract

Session C5: Multisensor Integrated Systems and Sensor Fusion Technologies

PEGASUS - Development of a Prototype, Self-trained, lorry Driver Coaching System Based on Geolocation, IoT and BI Techniques: Design Considerations and System Setup
Vassilis Gikas, Harris Perakis, Ioannis Stratakos, Panagiotis Sotiriou, School of Rural and Surveying Engineering, National Technical University of Athens, Greece; Dimitrios Pelekoudas, Fortion S.A., Greece
Location: Atrium Ballroom
Alternate Number 3

In recent years, the fast growth of ICT and IoT technologies has played a significant role in servicing the pressing needs for more efficient transport of goods. Particularly, the development of driver assistance and coaching systems aiming at improving safety at road and reducing CO2 emissions have seen a rapid expansion for heavy vehicles, mainly large trucks and transponders. Notwithstanding new, commercial driver coaching systems have shown up recently in the market, these systems have been designed to operate as black box devices for the user, while their functioning capabilities are still very limited. Given the enormous numbers of truck fleets currently operated worldwide, the necessity for an adaptable driver coaching system capable of providing real-time functionality is evident.
This study reports preliminary results from a research project aiming at developing a self-trained, driver-coaching system based on geolocation, IoT (Internet of Things) and BI (Business Intelligence) techniques. The system by design aims at reducing operational costs for the truck fleet, improving road safety, as well as contributing to green transportation. In effect, the system has been designed to produce user-friendly, two-way advice to lorry drivers tailored to the road (e.g., geometry, road markings) and operational (e.g., weather, traffic) conditions in real time and in a dynamic manner.
The proposed system is relies on: (i) a set of interconnected, heterogeneous high / low cost sensors and IoT data management tools; (ii) geolocation techniques used to determine the vehicle's current kinematic state, and (iii) analyses of business intelligence observations for calculating a set of Key Performance Indicators (KPIs) of the vehicle status bearing fully parameterization and extension capabilities. The methodology is developed in three main areas: (i) the design and development of the IoT network used for interconnecting sensors, (ii) the development of a customized BI model, and (iii) the implementation of geo-location principles and techniques required to fulfil the scope of the research. More specifically, the role of each of the three components of the system discussed in the paper are as follows.
(i) The IoT network features a multitude of on-board sensors, which collect data to describe the vehicles’ kinematic state - primarily, position, speed, acceleration and orientation. The data are collected using an On-Board Data Hub (OBDH), which serves a dual role; firstly, to send data to the BI system, and secondly to compare in real-time the vehicle's kinematic state with the rules defined within the BI system leading to the generation of appropriate audio feedback to the drivers.
(ii) BI is typically used to analyze large volumes of data for strategic decision-making at a commercial and corporate level. In the proposed implementation, a large amount of data is analyzed by OBDH to help in decision making (e.g. gear change, braking, acceleration). The BI system collects sensor data from the vehicle to produce a multi-dimensional matrix leading to the optimal voice command choice for each of the relevant metrics that operate as matrix dimensions.
(iii) The inclusion of geo-location data into the KPIs calculation system is a significant improvement over other existing “driver coaching” systems, as it takes into account the vehicle condition (e.g., road curvature, uphill / downhill movement) on the road.



Previous Abstract Return to Session C5 Next Abstract