114 Progress toward Cyberimmortality ing, the onboard computer vision system detected that the spacecraft was moving to one side more rapidly than desired, so  the  control  system  immediately  compensated.  Because there was no human being on board, and the speed-of-light limitation prevented information and commands from going to and from Earth in time, autonomous vision was crucial. But the robot vision system was not capable of recognizing objects, merely measuring the speed that objects and textures were moving across the field of view. Computer vision is pro- gressing rapidly, but it has a long way to go before it matches the capabilities of a sparrow, let alone a human being. Similarly, the computer techniques called natural language processing (NLP) are progressing rapidly. [23] Imperfect but serviceable speech recognition programs now operate many companies’  telephone  information  and  reservation  systems, and  the  quality  of  this  technology  improves  constantly. Perhaps the greatest controversy in NLP right now is how much  can  be  accomplished  by  brute-force  statistical  sys- tems without incorporating the results of linguistics research about the structure of language and the meaning of words. For example, a speech recognition system may consist of a set of mathematical models of phonemes – the individual sounds that comprise speech when strung together – and a statisti- cal model of the probability that various words would appear together in a sentence, based on analysis of a vast corpus of written language using techniques such as Hidden Markov Models (HMM). Some researchers are beginning to add lin- guistic rules, word definitions, and information from other modes of communication such as facial expressions. But NLP is not yet a fully ‘intelligent’ system that could take over a person’s speaking and listening functions, let alone thought processes.