Simultaneous Localization and Mapping (SLAM) is studied in the past decades, resulting in formulations that allow robots to estimate their pose and model their environments. In the SLAM framework, a robot does not care about planning, so that a man gives the robot the trajectory to follow during the SLAM process. On the other hand, planning algorithm is studied in the past decades and they assume that all sensor data is given, which means that the robot already knows the entire environment and is able to localize itself accurately. However, in real scenario, such as planetary exploration, search and rescue activities, and underwater inspection, it is critically important for a robot to have a fully autonomous navigation capability under uncertainty. Thus, we need an approach that can deal with the SLAM and planning to have autonomous navigation in the unknown environment.
The purpose of the Simultaneous Planning Localization and Mapping (SPLAM) is to have robot take an action that results in improved SLAM performance under uncertainty. Currently, I am thinking of using Model Predictive Control (MPC) methods to solve this.