Simple robotics controller which are able to follow a line can be realized without much effort. A well known strategy like a behavior tree can determine the next action of the robot and this allows the device to fulfill a task. More complicated challenges will show the limits of this strategy. If the aim is to plan a longer horizon with different subgoals, a new kind of control strategy is needed.
In the literature such a hierarchical planning system is called a task and motion planner. It consists of lowlevel planning with skills and high-level planning which are realized as PDDL symbolic representation. Often, such systems are equipped with Question&answering capabilities in natural language.
- WANG, Zi, et al. Active model learning and diverse action sampling for task and motion planning. In: 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2018. S. 4107-4114.
- Gordon, Daniel, Dieter Fox, and Ali Farhadi. "What Should I Do Now? Marrying Reinforcement Learning and Symbolic Planning." arXiv preprint arXiv:1901.01492 (2019).