From Scholarpedia
Jump to: navigation, search

    I would recommend talking about why the physical body is important when the goal is to try to get at foundational artificial intelligence. More generally, the question is how does an organism solve its problems. It does not have to be all computational. Other than that, the article is short and very excellent.


    2nd review: There is a fine selection of real-world examples presented here, but none of these are fully in simulation. In practice, a significant proportion of work in this field is done in simulation, but apart from the opening "An animat is a simulated animal or a robot ...", and the clue from the name of SAB conferences, the casual reader might not appreciate this simulation aspect. I would propose adding one or more examples of animat work done wholly in simulation; plus a reference to the issues raised when people question the validity or naivety of such simulations -- how realistic do they need to be? Otherwise, I agree with the first reviewer that the article is short and excellent.


    The article is well-structured and easy to read. I appreciated in particular how the authors succeeded in describing the main points in such a synthetic way. The article, however, can be improved by implementing the following minor revisions:

    To allow them to "survive" or fulfill their mission in unpredictable environments, the control >architectures of animats often implement, beyond mere reflexes connecting their sensors to their >actuators, a variety of other adaptive properties or processes such as active perception, efferent >copies, motivations, emotions and action selection. Often, these properties or processes serve to cope with present circumstances only. This is why more-cognitive architectures, able to deal with past and future events as well, are endowed with >memory and planning capacities.

    I suggest distinguishing more clearly between characteristics or properties of the controller (e.g. efferent copies), and capacities or processes (active perception, action selection) that the animat should have. Moreover, I would avoid using terms such us “motivation” without defining them. Also I do not see the connection between the term “adaptive” and some of these characteristics listed (e.g. in which sense efferent connections are adaptive, or for example, should the animat necessarily be provided with an action selection capabilities which change/adapt over time ?). I suggest to rephrase this sentence. For example, you might say that an animat might need to posses not only simple reflexes but also more complex capacities such us the capacity to exhibit a variety of different behaviors, to take into account previously experienced sensory states, and to regulate its behavior on the basis of the current environenmental conditions, etc.

    Moreover, other adaptive mechanisms, like learning, evolution and development, are also currently incorporated into them.

    Please rephrase. Saying that the evolutionary process is incorporated in the animat looks misleading to me. Here you might also include a pointer to the scholarpedia pages on developmental robotics and evolutionary robotics.

    More specifically, at the end of the project, this artificial rat should be able to exhibit the following adaptive capacities:

    I would avoid including research plans in an article of this type and I do not see the reason in this specific case given that the ICEA project is now toward its end. I rather suggest to include a brief descprition of some of the most significant results achieved, possibly by providing references to articles that can provide more details.

    Personal tools

    Focal areas