If the torpedo works so well in the seas why not using it on land? The answer lies in the mud and craters of no man's land, that would have made the travels of these little cars very difficult and probably too short.
This notwithstanding the invention of umanned veichles remains one of the highest aspiration of World War I technological imagination, especially in the U.S. . The vehicle, patented in different versions, had to carry at least one ton of explosive. Its detonation was controlled through a wire, by a timer or by a contact sensor. The wires connected to the machines could, according to the patents, control the car's speed (but not its direction: only in a straight line) and even be used to recover the torpedoes that missed the target. The machine was meant to replace the artillery, solving the (all to real) problem of the shell barrage alerting the defenders of the immininent infantry attack. Mass production was thought to be cheap: between 150 and 1000 dollars a piece, whereas a sea torpedo costed 7000. To reduce costs it was even proposed to use second-hand cars.
The sea metaphor is even closer in the case of the “underground submarine” or “subterrane” according the definition given by its inventor. In reality the machine was similar to a sea torpedo: unmanned, it was supposed to dig its way to the opposing trench, by excavating the soil in front of itself and releasing it from the rear. Its travel beneath no man’s land could last weeks, it was said, but the 120 cubic metres of dynamite that it carried would have guaranteed the destruction of the targe
 H. R. Everett, Unmanned Systems of World Wars I and II, (Cambridge, Massachusetts: The MIT Press, 2015)