Automotive Applications

Automotive Applications

Parking assistance system detecting pedestrians and surrounding vehicles
Parking assistance system detecting pedestrians and surrounding vehicles

According to the World Health Organization, over 1.2 million people die in road traffic accidents every year and approximately additional 20-50 million sustain injuries of different degrees. In fact, the majority of accidents is caused by mistakes committed by the human driver.

Many existing driver assistance systems such as Anti-Block-System (ABS) and Electronic Stability Program (ESP) provide basic assistance for the control of the vehicle in critical situations. Thus, these systems play a crucial role in modern cars, protecting the driver and the passengers.

Due to the increasing complexity of traffic, we seek for advanced driver assistance systems (ADAS) that are able to act according to the environment of the vehicle. Typically, ADAS require the evaluation of huge amounts of data captured by different sensor systems, involving radar, ultrasonic, video, lidar etc. Consequently, efficient and adaptive Signal Processing combined with Machine Learning are key elements for the success of those systems.

The Signal Processing Group is collaborating with the leading automotive industry, e.g. Adam Opel AG and Continental GmbH, which provide expertise knowledge and real data experiments for automotive applications. Research for automotive applications at the SPG involves

  • Regression and Classification
  • Decision Making
  • Radar Signal Processing
  • Detection
  • Data Fusion

For more information on Automotive Applications, see the sections below or contact the respective Research Associates.

Projects

Simulation of a vehicle equipped with several senor systems overtaking a truck
Simulation of a vehicle equipped with several senor systems overtaking a truck

Advanced driver assistance systems must be able to make correct decisions based on the information provided by the sensor systems mounted on the vehicle. While the general question as to how to learn to make decisions has been already solved for a wide range of applications, two major issues remain to be solved: how to learn efficiently and how to make decisions reliably?

Efficiency is of high importance for automotive applications as, eventually, we deal with a real-time system that must be able to make a decision within a restricted period of time. Further, when designing an ADAS, we need to make sure that the resulting system makes correct decisions – even if the sensor data is corrupted up to a certain level.

In this project, we seek for methods for situation assessment and planning that can be applied to automotive applications. For this purpose, we borrow concepts from Machine Learning to derive solutions with the requested properties.

For more information on this research project please contact Abdelhak Zoubir.