Robot Mapping

Abstract

Robot mapping is concerned with developing techniques that enable a mobile robot to construct and maintain a model of its environment based on spatial information gathered over time. Typically, the spatial information stems from directly perceiving the environment through external sensors. In addition, internal sensors like odometry provide information about change of location within the environment. There are, however, many more ways of acquiring spatial information, including external representations such as floor plans, sketches, or written descriptions, as well as direct communication with other robots or with humans.

Most approaches to robot mapping are incremental in the sense that a new observation is used to adjust the current spatial model, leading to a new model. The observation is then discarded. The adjustment of the model when new information about the environment becomes available can be seen as a two-step process: First, in the localization step corresponding features contained in the new information and in the current model are identified (data association) and the robot’s position within the model is updated based on the found correspondences (position update). Second, in the map merging step the spatial information in the model is complemented and updated based on the results of the localization step. The information flow of this general incremental mapping cycle is illustrated in Fig. 2.1. The current spatial model serves as input for the data association and map merging steps and is modified by the map merging step.