• driving
• embedded
• Paolo Andreucci
• PERCRO
• Vehicles

Nowadays in the automotive sector we are observing an increasing number of electronic devices embedded into vehicles and for this reason more and more data information are becoming available. So automatic driver’s profiling and assistance devices are becoming more and more relevant in the automotive market for different reasons.
First of all they can be used to detect incorrect driving behaviors improving driver’s attitude and hence the road safety in general.
Second by improving drivers attitude they can also improve vehicle related performance such as fuel consumption, tires usage, brakes usage and so on. Another perspective on which the driving profiling arouses interest is surely the motor sport world, in this context driving profiling will be used to improve driving performances instead.
Last but not least in past years driving profiling has attracted the interest of insurance companies thanks to the introduction of the Pay-As-You-Drive (PAYD) approach. This approach offers drivers the option of having premium that fit their individual driving patterns.

The objective of this thesis has been the development of a new driver’s assistance device which, based on real-time acquisition through embedded sensors and data analysis, can provide both driving assistance for drivers and technological support for driver’s profiling.
The work was initially originated from a first idea of Paolo Andreucci, the eleven times Italian Rally champion, who has developed the first satellite navigator mobile application, that alerts the driver of the difficulty of any turn with enough advance to drive much more effective and performing.

The proposed approach is based on the monitoring of driving behavior through the definition and automatic extraction of suitable algorithmic features, driven by relevant events associated to driving performance such as brakings, shifts, turns and overtakes.
Such features represent the starting point of drive evaluation activity which have been implemented using fuzzy system.

The development of this work has required a substantial effort, first to assess whether it was possible to estimate and detect incorrect driving behaviors through a low-cost device composed by IMUs, GPS module and an OBD device. This has required first the definition of incorrect driving behaviors and the definition of relevant numerical parameters associated to them, then the mapping to suitable data measurements acquired by on-board sensors.
Second it has required the development of a driving model, experimental acquisitions in relevant driving scenario conditions and validation of such hypothesis through acquisition tests and data analysis. Acquired tests data have been analyzed and processed to extracts features used in the last phase, the drive evaluation. Additional tests have been performed to analyze and understand parameters with which it is possible to classify and distinguish different driving features. Such parameters have been used to build the drive evaluation fuzzy system.