Project description :
Summary (MAX of 2100 characters with spaces)
This project aims to establish new design methods for robust and efficient automotive navigation and optimal management of a fleet of vehicles in harsh environments. In addition, the project also aims to develop innovative metrics for real-time analysis of dangerous driving behaviour as well as real-time analysis of car accidents in order to significantly improve global safety of Canadian drivers. In general, this research proposes to combine measurements from a high sensitivity GPS receiver with data coming from a self-contained inertial navigation system and other complementary autonomous sensors such as odometers and magnetometers. Moreover, in order to provide an affordable solution, the targeted system will be based exclusively on the use of very low cost sensors. It is expected that this project will help reduce the environmental footprint of motor vehicles in addition to having a significant positive impact on overall vehicle safety. For example, improving vehicle localisation accuracy and robustness in harsh environments can significantly reduce the time to find a stolen or misplaced vehicle, which can have an important impact on Canadian companies’ finances. Furthermore, having a robust and precise solution for monitoring vehicle behaviour can lead to the implementation of a new taxation system based on car usage or on driving behaviour, which according to recent studies, can help reduce vehicle greenhouse gas emissions by up to 10%. In addition, accurate reconstruction of car accidents in real-time allow prediction of specific parameters of an accident scene thus improving reaction time and vehicle safety. The proof-of-concept demonstrator will be evaluated in-laboratory and on-road using simulation equipment and a car test platform under real operating conditions in order to characterize protocols and system performance. The project will contribute to international initiatives for the definition of new standards and contribute to Canadian efforts to reduce greenhouse gas emissions, and create new employment opportunities for the team of highly qualified personnel.
Responsibilities of the candidate:
According to the schedule, this master’s student will be in charge of the following tasks:
1) 3-6 Initial development of the advanced adaptive navigation model
2) 3-61 Mathematical modeling of the advanced adaptive navigation model
3) 3-62 Simulation of the adaptive navigation models
4) 3-63 Tests and validation of the simulated adaptive navigation model
5) 4-11 Integration of calibration models to navigation systems
6) 4-12 Integration of system constraints to navigation systems
7) 4-13 Results analysis and systems comparison
8) 4-14 System weakness identification
9) 4-15 Study of possible architectural improvements
10) 4-16 Validation / performance analysis of improved system
The general objective of the Master’s research project is to study an adaptive sensor fusion scheme based on neural networks, in order to implement it into a multi-sensor integrated navigation system for automotive applications. The student will begin his master’s study based on the work of the Ph.D. student regarding the in-depth theoretical background. Assisted by the Ph.D. student, he will first realize the mathematical modeling of the neural network in order to develop a completely adaptive navigation system based on a multi-sensor fusion approach. Following this, he will achieve the implementation of the model in Matlab / Simulink. This algorithm will then be tested intensively using simulated measurements and validated on real sensor’s measurements. After this validation, he will assist the Ph.D. student on the real-time implementation of the algorithm into the Orchid VTADS prototype. Finally, the student will conduct an intensive series of tests in realistic automotive scenarios in order to assess the performance of the developed system compared to alternative implementations realized by other students in the research team.
