Molecular Manufacturing and the Private Aircar

J. Storrs Hall, PhD

Earlier this year I spent some time studying the problem of the design of flying cars.  This was initiated by a suggestion of Al Globus of NASA Ames, who thought it would be a good way to get some of the aeronautics people at Ames to appreciate the possibilities of  molecular manufacturing.  I'm very grateful for the help of Eric Drexler with the project, and also some suggestions from Jeffrey Soreff.
Surprisingly enough, it's a non-trivial problem even with molecular manufacturing. For fifty years, the private aircar, i.e. an aeronautical vehicle which is functionally and economically capable of replacing the automobile, has seemed tantalizingly close to reality. Here's a montage of a few real, experimental, and fictional ones.  This hope has foundered on several stumbling blocks:

Performance Parameters

In considering the range of possible modes of transportation, there are two very clear local optima among all the possibilities that it seemed extremely desirable to match.  First is the family car, particularly in terms of its payload capacity and the convenience of operation it provides.  Second is the flying regime of commercial jets.  This is perhaps less well known.  For the 50 years preceding 1970, typical commercial air travel speeds increased exponentially; for the next thirty, they remained constant.  Airliners fly just above the tropopause and just below the speed of sound.

Thus our design goal is a machine that fits in your garage, takes off from your driveway -- quietly, isn't apt to flip over and dive into your house while doing so, and cruises at 6+ miles up at 500+ mph.
Designing such a machine illuminates many of the considerations that will in all likelihood be typical of many early products of molecular manufacturing.  First, the interface of the nano-mechanism to the macroscopic world is problematical.  Surface films of moisture affect exposed nanomechanisms like pouring molasses over conventional machines.  Secondly, the major considerations in conventional vehicle design, having to do with sufficient power in the motor and weight (of motor and vehicle), essentially disappear.
One of the most remarkable figures calculated in Nanosystems is the power-to-weight ratio of the electric motor described in section 11.7.  For comparison, a typical automobile engine might produce 100 kW power.  Drexler motors producing the same power would occupy a volume of a tenth of a cubic millimeter, about the same as a single hair from one of my eyebrows.  In practical terms, this means that in molecular manufacturing designs, we can put motors everywhere with virtually no cost in weight or volume.
This ability does not come without a cost, however.  The greatest cost to the designer is control complexity.  This is to some extent the same problem that has come to be the bane of software; it is easy to create systems so complex that their designers do not fully understand them, to the detriment of, among other things, their reliability.  Thus it behooves us as designers to exercise some restraint.

Shape Changing

The quietest way to take off is to jump.  Furthermore, if you're taking off in an area with trees and buildings, and/or gusty winds, it is advantageous to maintain contact with the ground until you're above the height of the obstacles.  Thus our aircar will be equipped with extensible legs for jump-takeoff.  I've designed legs that weigh less than an ounce, fold up to be less than an inch long, and extend to the height of a five-story building. They're essentially telescoping cylinders of diamondoid.  They contain thousands of motors, but in keeping with the principle of simple control, all the motors do the same thing at the same time; the leg as a whole has one degree of freedom.
Shape-changing is virtually certain to be a major feature of molecular manufacturing products. One of the more important contributions of this work and my previous work on Utility Fog has been the development of a general scheme for shape-changing based on the theory of laminar fluid flow.  Laminar flow, as described by the Navier-Stokes equations, is fairly well understood, at least as compared to other ways methods of describing a phenomenon of similar complexity (such as turbulent flow or sets of specific routing instructions to each part of a shape-changing mechanism.  The novel part of the work is to develop a discrete version that is applicable to multitudes of machines instead of a continuous fluid.
If a reasonably well-characterized range of shape changes is desired, such as the changes in the size and shape of a wing that adapt it to different regimes of flight, we can pre-calculate the flows and produce a mechanism made largely from very thin diamondoid sheets that slide across each other (powered -- millions of Drexler motors embedded in them).  This enables us to design a machine whose wings resemble those of a bird for flying low and slow, those of an airliner for flying high and fast, and disappear completely when on the ground. Careful design allows us to produce structures with a substantial fraction of the strength/weight properties of rigid ones.  Thus it seems quite achievable to have a vehicle with a tare weight of 100 kg or so.
Another major advantage of shape changing for aircraft is that in flight, all unnecessary gadgets, legs, appendages, and other protuberances can disappear completely. leaving a clean. completely streamlined shape (and greatly reducing drag).  Indeed, we can steer and control attitude by means of shape changes in the wings and body, meaning that we don't have to have tail fins, canards, or other separate control surfaces, further reducing drag.
(Note, by the way, that using these techniques it is quite straightforward to design a robot of humanoid height and strength what weighs 5 grams and could collapse to the size of a ball-point pen (which incidentally weighs 5 grams).  Note also that programming the controller for such a robot is a considerably more difficult task, if indeed it is possible.)


Extensible legs.  The structure supports the weight by itself but is pressurized to prevent buckling.

Quiet Takeoff

The problem of noisy takeoff remains.  The physical parameters of the problem are that the noise depends on the velocity of the stream of air thrown downwards.  For reaction forces in the 10,000 Newton range, (assuming vehicle, fuel, and cargo can weigh a tonne), any narrow stream, such as a jet or rocket, will be too noisy regardless of how it is produced.  For typical parameters we need a stream cross section of at least 10 square meters (e.g. a 12-foot circle).  This cannot, however, be produced by a propellor or rotor blades, since they produce substantial noise by other mechanisms.  My proposal is a retractable sail of "fancloth" -- a fabric with a mesh about that of window screen (1.5 mm) where each space is occupied by a tiny fan.  The size of the fans is a tradeoff; if much smaller they are too easily choked by atmospheric dust, and if much larger they generate noise and turbulence.  (As it is they are operating at a very low Reynolds number, and thus can likely avoid creating turbulence; on the other hand they have increased viscous drag and thus consume a bit more power.)  Note that although MEMS technology could build the fans today, no existing technology could provide the motors.
Taking off.  The blue "skirt" is the fancloth.

Safety, Maintenance, Etc.

It may not be generally known, but current computer hardware and software is capable of driving an automobile from coast to coast (e.g. CMU's "No Hands Across America").  Autopilots capable of flying the aircar with only indirect control from a human operator are eminently reasonable to expect within the next decade.  More problematical is the problem of airspace congestion should the aircar become popular.  Considerable work is needed in automatic distributed air traffic control; if the political obstacles can be overcome, the technical ones do not seem insurmountable.
The final aspect of the aircar that I addressed was the subject of self-repair and self-maintenance.  These are closely related to the subject of self-replicating and -extending machines, which is the core of my ongoing research agenda. (Pages on this will be forthcoming at a later date).