Autonomous Driving
Description
As a key emerging technology, autonomous (or machine) driving (AD) represents a key applied research area of our chair. Developing AD solutions requires a mindset of mechatronics, which goes far beyond its interpretation in mechanical engineering. Perspectives of control engineering, computer science and communication engineering offer versatality and dominate the mechanical domain as solutions climb towards higher levels of autonomy and connected driving. Our to-date AD solutions conjunct adaptive control algorithms with various modules for envinroment perception. Control algorithms are developed using not only the well established concepts such as model predictive control (MPC), but also the modern machine learning techniques such as IL, RL and DNN based stochastic MPC. The developed algorithms target various driving scenarios, including city driving, inter-city and highway driving, valet parking and more. In the scope of city driving, we also focus on connected driving for which collaborative and distributed decision making algorithms are needed. The developed algorithms are tested in real-world scenarios using two AD facilitated vehicles, which are equipped with high-end sensory and computation resources, as well as a software stack for perception, delivering accurate real-time information, preception and understanding of the environment.
For more details click here.
Keywords
Autonomous Driving
Environment Perception
Decision Making
Vehicle Control
Contact
Prof. Naim Bajcinca
Gottlieb-Daimler-Str. 42
67663, Kaiserslautern
Phone: +49 (0)631/205-3230
Fax: +49 (0)631/205-4201
Funding
BMVI, Bundesministerium für Verkehr und digitale Infrastruktur
Autonomous Driving
One of the basic features of any AD algorithm is lane or path following. Given a road with lane markings, the objective of the controller is to steer the vehicle along the lane while taking care of the speed limits, stop signs and also road obstacles. We achieve this by implementing a model predictive controller (MPC) based on a kinematic model of the vehicle. The choice of using a kinematic model for the vehicle dynamics is to keep the controller generic and agnostic of any vehicle type. This is made possible due to a robust MPC implementation which is also fast enough to correct errors occurring due to model misspecification. The perception algorithms include a library consisting of modules for object detection / tracking, free space detection, etc. In this context, we are particularly occupied with developing holistic solutions embedding machine vision and control theory. Further modules of such a holistic picture towards higher levels of autonomy address planing and decision making algorithms.
Architecture
