The vacuum cleaner robot is a common didactic example in the field of multi-agent systems. Normally, the robot is placed inside a grid world and reasons directly on the step to perform. In this work, we propose a different approach where the robot body and the robot mind are separated, the body is placed in a continuous environment and the mind reasons in a more abstract way on regions instead of cells, leaving to the body the task of moving continuously. In this way, the robot is able to abstract more and focus on high-level tasks instead of low-level details. An implementation of the proposed approach is presented with JaCaMo agents and a Godot simulation, using Region Connection Calculus (RCC) as base for the agent.