In order to effectively drive a remote ground vehicle using video images, the operator must be provided with a natural, real-time video sequence and the imagery must be accurate and detailed enough so that the operator can make mobility and survivability decisions. Unfortunately, high data rate communication channels are often not feasible for this task. To accomplish remote driving using a low data rate channel, video compression techniques must be incorporated. This paper discusses the remote vehicle driving problem and describes several video compression algorithms that have been implemented on PIPE, a real-time pipelined image processing machine. The paper then discusses how these algorithms are evaluated on real-world remote driving tests. Finally, advanced techniques for video compression are proposed.