In this paper a creative action planning algorithm (CAPA) is presented for solving multiagent planning problems and task allocation. The distributed multiagent system taken in consideration is a system of m autonomous agents. Agents workspace contains simplified blocks which form different space structures. By employing the planning algorithm and through interaction agents allocate tasks which they execute in order to assemble the required space structure. The planning algorithm is based on an inductive engine. From a given set of objects which can differ from the initial set agents need to reach a solution in the anticipated search space. A multiagent framework for autonomous planning is developed and implemented on an actual robotic system consisting of three 6 DOF industrial robots.