In this task the consortium will attack the algorithmic aspects of controller synthesis and planning including approaches that prefer a simpler sub-optimal strategy over an optimal one. Modification and extension of heuristic methods from model checking and operations research will be investigated to allow for transfer to (optimal) control synthesis and to the closely related domain of online scheduling problems.
In analysing timed automata models obtained from PDDL+ domain descriptions, the consortium expects to benefit from static analysis of the PDDL+ source in obtaining information about independencies and mutual exclusions between action-instances. A specific structural property of timed automata models of job shop scheduling problems is that they are acyclic. This reduces their complexity and provides for specialised algorithms for analysis and synthesis including heuristic forward search, especially in the context of optimisation.