Simulation and planning with multiple representations

Presents a research work in the field of robotics. The system described uses a combination of analogical, qualitative and symbolic knowledge representation formalisms in order to plan the right actions in assembly situations. The system applies a simulation of physical objects and phenomena based on analogical representation and bottom-up description. In order to extract information relevant to the status of the simulation, an interpreter is included that allows the problem solving system to understand the result of each action. The cooperation between planner, simulator and interpreter allows the system to recognize whether there is a progress toward a correct solution or toward a failure