Abstract
Robots and agents are both autonomously interacting with their environment. It is even argued that robots can be regarded as embodied agents. This paper discusses the use of Agent Oriented Techniques for controlling robots, their benefits and their limitations. Basic skills need much more emphasis for robots than related skills of agents. Algorithms for perception in the real world are a key problem for robot programming with many (still unsolved) challenges. More deliberative tasks like planning can benefit from techniques developed for agents. Since real environments are highly dynamic, classical planning techniques are not flexible enough, but there exist more sophisticated techniques for agents which are useful for robotics as well.