Spratt.Multics@MIT-MULTICS@sri-unix (09/10/82)
There was a project in the Man-Machine Systems Laboratory at MIT around 1977 to build an remote-controlled underwater exploration vehicle. The problem here was that the communications with the vehicle (which had a TV camera mounted on it) was done via sonar (very narrow bandwidth communications). To get a picture transmitted required about 8 seconds. Well, actually, the total time delay on dealing with the vehicle was 8 seconds, but transmission of the picture was the bulk of the delay. The approach in this project was not to put intelligence in the vehicle, but have a detailed model of the world of the vehicle which was used to predict the effects of operator decisions. The graphic display to the operator depicted the last actual picture and the predicted modifcations in a fashion which made it possible for the operator to easily tell which was which. Someone must still be doing this sort of work, and it seems like it would be readily applicable to the extraterrestrial remote-control problem.