One might wonder what Şahin et al. are trying to afford, but this is not a paper on robotic affordances. Instead, the authors try to reason whether to enter a heated debate--many dilemmas and criticisms might arise from reading this paper. Nevertheless, it is worth reading, regardless of the school of thought to which you (can afford to) subscribe.
Based on the concept of affordances as defined by Gibson, the mid-20th century psychologist, Şahin et al. provide us with a superframework for robotic affordances. In layman’s terms, affordances in this context refer to the concept of whether a robot decides if an action in a given situation will work for it, and which action to take--based on its interaction with its current environment.
Şahin’s robots do well, given the environment in which the framework was simulated--but that is not what this paper is really about. This paper is about the generalization of Gibson’s view on the term affordance. They claim that not one but three views exist on this term: environmental, agent’s, and observer’s perspective. That is the reason for the controversial discussion of this subject across scientific disciplines, from psychology to robotics.
This is a well-written paper that’s easy to follow and read; it contains a little bit of theory mixed with a few experiments. If the formalisms are too difficult to follow, there is plenty of discussion to fall back on.