Realtime Decision-Making for Certifiable Autonomous Unmanned Systems
Jens Halbig, André Windisch, Patrick Kingsbury, Norbert Oswald, Wolfram Hardt · International journal of imaging and robotics · 2013
Today the trend in aerospace and automotive sector migrates from the piloted vehicles to the so-called unmanned systems. To realize the autonomous behaviour of these systems, it is necessary to integrate decision-making and planning algorithms. The theory of these algorithms are well known from artificial intelligence and robotic research areas, but currently there are no realtime implementations available that fulfill the requirements of modern avionic and automotive software architectures. These architectures are based on certifiable time-triggered embedded realtime operating systems which assume an a-priori calculation of time and space bounds of all the system components as a prerequisite for their integration. This paper presents a new approach for realtime decision-making which fulfills these requirements.