Multi-Modal Sensor Based Localization And Control Method For Human-Following Outdoor Security Mobile Robot

Recently, mobile robots are attracting attention in various industries. Outdoor unmanned security mobile robot is a key issue for surveillance. The main task of these security robots is to protect people and property. For this purpose, the robot should be able to autonomously navigation and interaction with people is essential. Especially, it is required to perform autonomous driving by avoiding collision with humans and obstacles, tracking a certain human for intruder surveillance or safety of people. For the outdoor security task, we propose the novel localization and control methods that is not only overcome the weather conditions for positioning, but also build a safety route for avoiding and following relative to human position. The robust localization method is based on detecting the salient features by the information filters for multi-layered knowledge (e.g. sensory, episodic, semantic and cloud big data), and then estimate an accurately position of the moved robot. Next, the safety route is defined a rollover model of a security robot on slope and suggests a path generation using DWA (Dynamic Window Approach) method with safety ratio. The method of evaluating rollover is the ZMP (Zero Moment Point) concept. If there is a ZMP between the wheels of the steering mobile robot, it can be safe. The results show that the autonomous navigation is possible with robust localization method, and then it can follow a specific human by the safety path generation. The proposed method is expected to be usable in various applications requiring outdoor surveillance.