A robot simulation system integrated with a task planner and an Environment Memory Base

In this paper, a robot simulation system which can accomplish multi-task planning is proposed. The system is based on Simbad - an autonomous robot simulation package. Compared to Simbad, our system can establishes the environment model, and enables robots to make proper planning under uncertainties. The main contributions of this paper are 1) integrating the “Environment Memory Base” in the simulation system to describe the environment information, which stems from the simulated virtual environment; 2) integrating the task planner (HTN Planer) in the simulation system to decompose the main task into several sub-tasks; 3) proposing an approach using the Environment Memory Base and the task planner to complete a multiple task in the simulation.