Skip to Main content Skip to Navigation

Embedded multiprocessor architectures for automative driver assistance systems

Abstract : Automotive crashes are responsible for the highest number of accidental deaths all over the world. Researchers, automotive manufacturers and government authorities around the world are continuously looking for solutions to this problem. Research has shown that half of the accidents can be avoided if a driver is alerted to an impending collision a fraction of a second in advance. A mechanism for warning the driver of an approaching danger is called a Driver Assistance System (DAS). Accident statistics show that a great majority of the vehicle crashes result from front-end collisions. Hence minimizing frontal collisions would significantly decrease road accidents. To predict a front-end collision sufficiently in advance, the obstacle must be detected from a distance. Moreover, for the DAS to be really effective, an imminent collision must be sensed in all circumstances, especially in poor weather where the DAS is needed most. A radar sensor fulfils both the prerequisites of long range obstacle detection and all-weather operation. However, only detecting obstacles can be useful to a certain extent. To establish whether an obstacle is on a collision course with the host vehicle, its trajectory must be foreseen before it comes close to the host vehicle. Determining the trajectory of a moving object requires its dynamic behavior to be monitored over a period of time. In a real traffic scenario more than one obstacle can pose danger to the host vehicle, hence trajectories of multiple objects have to be monitored simultaneously. An apparatus which is capable of performing such functions is called a Multiple Target Tracking (MTT) system. In this thesis we propose a DAS using the principles of Multiple Target Tracking to monitor the dynamics of obstacles hundreds of meters ahead and to avoid a collision of the host vehicle with them. While theoretically such a system offers one of the best answers to the road accident problem, its practical implementation is not a trivial task. It involves complex computations and consequently, needs a long processing time. However, to alert a driver to an approaching danger in real time, the computations must be performed very rapidly. We use multiple processors in our system to share the computation load and thereby reduce the processing time. Multiple processors running in parallel not only speed up the computation but also address the power consumption issues of the embedded systems. We use FPGA (Field Programmable Gate Array) as the implementation platform for our multiprocessor system. FPGAs offer the flexibility needed for the ever evolving embedded systems and they are very cost effective. A multiprocessor system implemented in an FPGA makes its architecture flexible and reconfigurable while the processors can be reprogrammed when needed. Thus FPGA based multiprocessor systems guarantee flexibility in hardware as well as in software therefore they scale very easily. We optimize the system architecture to minimize its hardware size while still meeting the realtime deadlines of the application. Minimized hardware not only leads to reducing energy consumption of the system but also enables us to fit the system in a smaller FPGA which plays an important role in reducing the cost of the system.
Complete list of metadatas
Contributor : Julie Cagniard <>
Submitted on : Monday, December 14, 2020 - 5:27:48 PM
Last modification on : Thursday, December 17, 2020 - 3:32:43 AM


Files produced by the author(s)


  • HAL Id : tel-03065222, version 1



Jehangir Khan. Embedded multiprocessor architectures for automative driver assistance systems. Embedded Systems. Université de Valenciennes et du Hainaut Cambraisis, 2009. English. ⟨NNT : 2009VALE0034⟩. ⟨tel-03065222⟩



Record views


Files downloads