Simultaneous Localization and Mapping with Basic Cognitive Understanding of Environments

摘要

Simultaneous Localization and Mapping (SLAM) is a well-known algorithm in autonomous mobile robotics research. For the current classic SLAM algorithm, it allows a mobile robot to localize itself and at the same time build a map according to the layout of the environment. But the map that builds by SLAM provides only geometry information and without any cognitive understanding of the environment. Cognitive understanding of the environment is necessary because when a man navigates in an unknown area, he needs to recognize the objects around the environment and interpret the environment to have a better understanding of his current location. For example, if a man was kidnapped to an unknown location, after removing his blindfold, he would able to know that he is in a warehouse by recognizing the surroundings objects like racks, boxes, and goods. Therefore, we propose Cognitive SLAM (C-SLAM) that enables mobile robots to classify and differentiate the unknown environments through multiple objects recognition. One of the potential application examples is vacuum robots. With C-SLAM, a vacuum robot will be able to classify the living room, bedroom, or dining room, and then able to schedule to clean the dining room more frequently than the bedroom.