Semantic Information-based Reliable Autonomous Navigation in Wide Space

Recently, much attention has been paid to intelligent robots. Especially, autonomous navigation of robots is the most important technology and is being developed by many researchers. The autonomous navigation technology is based on the SLAM to find the position from the sensor data that is mounted robots. However, most methods find locations based on high-performance sensors in the predefined environments. This is difficult to apply to complex environments such as wide area due to the difference in locomotion and sensor performance. Therefore, efforts should be made to improve the application of limited navigation technology. In this paper, we propose a method to drive in a wide space for robots with various locomotion and sensor by semantic information based autonomous navigation method. By using semantic information, the robot recognizes the surroundings using available sensor data and performs autonomous travel. For this purpose, a semantic map for a unit space (e.g. a room, a hallway, road etc.) is generated and traveled by receiving information suitable for a robot locomotion and sensor configuration from the local server. The proposed method utilizes the semantic map to drive in the same way as a person in a large space, and can use intelligent robot driving using the property information of the object. Therefore, it is expected that industrialization of robot autonomous navigation will be promoted.