While the topic of human-safe robots has an appreciable academic history, only today are various solutions to this challenge becoming economically relevant. We are witnessing a maturing of technology and market expectations, paired with initiatives in the area of standardization. One of the remaining hurdles to routine deployment of robots for collaborative operation with humans is obtaining a well-founded understanding of low-level injuries of the type that such robots can potentially inflict. For the assessment of collaborative robot designs and systems, the low-level injury potential must be unambiguously quantified. A classification scheme for holding such information is presented and an outlook on its use is given.