For transportation companies, the break-even point for a truck is depending principally on the fuel price, personal costs and the use time. The InnoLog project presented here tries to automate the work yards management in order to decrease the personal costs on these places. A graph model of the work place was generated to define the different possible sequences. At each new registration of a truck, the subset of the free sequences was extracted to compute the shortest way between the different required stations. After that the truck scheduling was fixed with theses sequences. If there is no solution (lack of free sequences or time slot), the new truck contacted the conflicting vehicles to find locally a solution by postponing one of their actions. If no positive reply is found, they will themselves look for modifications of their planning in order to get a new time slot, depending on the priority of the tasks and the deadlines. To the end vehicles with lowest priorities will have to change automatically their schedules or be send to parking lots by using this recursive local optimization. Hence the complexity of the planning will be finally asymptotically linear which able to manage a massive amount of vehicles in real time