The article presents the Developmental Robotics field in a very clear, relevant and fair manner. It asks the right questions and give a good overview of this on going research.
I have just a few minor comments/corrections
- As Figure 2 presents the Infanoid robot, it might be a good idea to cite one of Kozima's papers in the list of work you cite as representative of experiments with real robots - "Core Knowledge: An organism cannot develop without some built-in ability". Such a general sentence is certainly true. However, Spelke's views are not uncontroversial or universally accepted in Developmental Robotics. Therefore I would turn the paragraph a bit differently or add additional references. Here, it seems that Spelke is one of the pillar of the field. - "Behavorial drives". I would not use this term which has a rather precise meaning. In the 1950s there were a big debate on whether such things as curiosity, creativity,etc were NOT drives, in the classic sense. I would use "Motivation / Motivational factors" or "Intrinsic Motivation" if you want to be more precise. - In the references, "Kaplan, F. Oudeyer, P-Y, Hafner, V.V. (2007)" is in fact "Oudeyer, P-Y., Kaplan, F. and Hafner. V.V (2007)"
The paper presents a brief but detailed overview of the field of Developmental Robotics.
I think the article is well written and is a useful reference for researchers who want to approach/gain more knowledge on the field. I think the paper is good as it is, here are a couple of suggestions that the author might want to consider to improve it.
I suggest to add some references to work related to Robotcub and Infanoid where Fig1 and Fig2 are first introduced. Also, it could be nice to add a caption to explain what kind of research has been addressed on those platforms. As it is the pictures seem just examples of "fancy platforms" but it is not clear what is their role in the field.
In the Challenges Section:
Self-exploration here seems a bit misleading. It reminds me more of mechanisms related to autonomoous learning (autonomous exploration of the environment or the robot's self). Starting the paragraph with self-programming could be more appropriate.
Learning causality. It looks to me that "active learning" is a boarder area, which include learning causality. Also the paragraph describes more the challenges related to the former issue, as opposed to the latter. I suggest either to name this paragraph "active learning" or explain better why you think learning causality is an important challenge.