Computer vision is an instance of the “if it’s easy for a human, it’s hard for a computer” truism. Seeing is so natural to us that we underestimate how hard it is. A significant portion of our brain is devoted to processing the signals from our eyes- and we only understand a small part of those brain structures. Computer vision started out in the 60s trying to identify the location and orientation of parts in manufacturing so that a robot effector could pick them up; that led to a set of techniques referred to as blob analysis. Model-based vision techniques try to recognize an object by its structural shape, but that can be difficult—for example, a hanging basket chair is as much a chair as a standard dining room chair with four legs. That led to more biologically inspired techniques where the computer attempted to recognize the functional properties of the objects, e.g., sittability, or the properties of being the right height and surface area to sit on. The biological methods also produced optic flow systems that can detect time to collision without recognizing what is the object the robot is about to hit. Face recognition or even recognizing a human remained difficult, with slow progress in indirect cues such as gait detection, which is described in Autonomous. There was little dangerous of a surveillance culture in The Robots of Gotham. But then Moore’s Law enabled artificial neural networks to have more than three layers and deep (as in many layers in the network, not deep as in intellectual or thoughtful) learning took off. Unfortunately, deep learning is reaching a plateau and the results appears to be very sensitive to the training examples-- so much so that science fiction now talks about nurseries where all the inputs to a robot or AI system can be controlled. But the real challenges are not object recognition but scene understanding and object permanence, topics explored in Kill Decision. Introduction to AI Robotics second edition (out in Sept 2019) is a good reference as well as the Springer Handbook of Robotics second edition.