Representations of the environment

All the sensors used in mobile robotics provide either information about the robot's movement (proprioceptive information), or information about the robot's environment (generally referred to simply as "perception").

This information has complementary properties. Proprioceptive information, which may come from measuring wheel rotation or acceleration using an inertial measurement system, is subject to cumulative noise which makes it unusable in the long term, as the error in position measurement is constantly increasing. For their part, perceptions, which may come from rangefinders or cameras for example, are subject to ambiguities which do not necessarily increase over time, but which do not allow a robot to be located unequivocally....