Guaranteeing Safety in Real-time Control Plans – An Approach in ORICA

T.A. Omar, A. Simonet, and M. Simonet (France)

Keywords

Realtime, artificial intelligence, realtime intelligentcontrol, planning.

Abstract

AI planning for the response of a Real-time control system is a process of reasoning about world states, based on possible incoming events and previous actions, and planning future actions that can lead the world to its goal safely. For hard real-time control applications, planning requires exact knowledge of the response deadline associated to each state. It must then guarantee that the planned action from a state will execute and finish-up before the failure deadline of the state arrives. In this paper, we present the problems associated with hard real time response planning such as the precise determination of the response deadline of a state and the allocation of the reaction time to the response actions to guarantee safety. The solution to these problems, as implemented in ORICA (OSIRIS Real-time Intelligent Control Architecture), is explained.

Important Links:



Go Back