Let's talk about Robothinking Robothinking is a word which represents the mix of Robotics and Artificial Intelligence, i.e, where the topics on Artificial Intelligence intercept the field of Robotics. Indeed, the robots of today are becoming increasingly smarter such that talk about robotics is really more of a talk about artificial intelligence and related topics. It's easy to find examples of robothinking. For instance, let's take that small vacuum cleaner robot that you bought to free your office of all dust. The most obvious task is: clear the dust. But, of course it, is expected that the robot autonomously move around the room to clean more than just the place where it was dropped in. Thus, most often the robot is wondering within itself, "Where should I go now?" The robot must be programmed in order to answer this kind of question. But the type of programming is not the type that we would use for programming a calculator to solve an arithmetic expression. Instead of handle a well shaped set of inputs and outputs, the robot must be programmed to LEARN the environment and MAKE DECISIONS under unobservable, incomplete, dynamic and/or noisily sensor readings. Having said that, robothinking is not all that different from how we human beings think. The robot will first need to perceive its environment. It could be as easy as reading an electrical switch or as complex as combining two or more realtime camera streams. In the case of the vacuum cleaner, perception is achieved from the use of infrared rays or ultrasonic pulses to detect objects and obstacles nearby. The next step is to infer (estimate/generate) a model for the environment. The model is a representation of the world of things where the robot live in or need to handle. Usually the model is “incomplete” because more often the sensors's readings do not cover the whole environment. Another source of incompleteness is when the environment is too complex to be modelled or, sometimes, the environment changes so quickly that the robot can just model a small slice of. The vacuum cleaner's model could be a map for the observed obstacles coordinates and distances. The vacuum cleaner robot does a strong assumption: things do not change over the time. Thus, the obstacles and elements in its map are always static. The “black regions” in the map are the regions where the robot sensor don't have data yet. It is the unobservable area. Once the robot has an environment’s model in mind, the next step is choose what to do.