nix wrote:Do you think it would work in space?
In short - No; and if it did, spacecraft designers would almost certainly already be doing it, and astronauts on space-walks would be able move themselves around with comical variations of my incredibly funky dancing (I may have got "comical" and "funky" the wrong way around there!)
In the first example, displacement is produced primarily because static friction ("stiction") is stronger than sliding friction. Ignoring inefficiencies of the pneumatics, energy lost to sound and heat, etc...
If the same amount of energy is going into the piston in each direction, moving it over a shorter time period creates greater force at the sliding surface (force = mass * acceleration), so stiction is easily overcome, and the "vehicle" part is then able to slide over the worktop as the lesser frictional reaction force is insufficient to oppose it, and the two parts separate from the barely moving centre of mass of the composite object.
Doing it over a longer time period creates insufficent force to overcome stiction between worktop and "vehicle", so the same energy is put into moving the "inertial mass" (and centre of mass of the composite object) by a larger distance. The analysis of rolling resistance is somewhat more complex than sliding friction, but is analogous enough in practice that the example with the bearings is essentially the same in principle.
Floating freely in the vacuum of space, there are none of the surface or rolling contacts necessary to create these conditions; so there would be no net acceleration of the centre of mass, just a redistribution of mass around that centre - the two piston movements would exactly cancel out regardless of the speed of movement, and the result would be the same as if you never actuated it. To make the vehicle accelerate and then keep moving in a vacuum, the mass moving in the other direction has to separate from the vehicle, because so long as they don't interact with anything else, the problem can always be reduced to a single rigid object with the same mass and centre of mass as the composite object. Even if they separate, the system as a whole is equivalent to them being joined with a massless piston rod that can be extended forever.
There is no way to do it without effectively requiring the vehicle to lose mass (e.g. the products of the burned fuel ejected from a rocket engine), at least, not without harnessing gravitational acceleration or some other external influence.
Sorry to be a party pooper! But kudos for your inventiveness - and I'm incredibly jealous of your shed!