Our idealized assembly process starts from a Computer-Aided Design (CAD) description of some object to be built. Computer-Aided Manufacturing (CAM) software decomposes the object into primitive building blocks and then into an assembly sequence. The assembler control computer uses this assembly sequence to control a huge number of nanomanipulators performing mechanochemistry to build the desired object. Each nanomanipulator may only be capable of moving a single molecular scale building block around, but since there are a huge number of these nanomanipulators working simultaneously, the system is capable of building a large quantity of products.
The initial assembler will not include an onboard computer or power source for the manipulator, so must be controlled and powered from outside the device.
The system design required for a practical assembler must deal with how to get power and signal into a huge number of nanomanipulators, as well as how to get feedstock to them and take finished materials away after construction. Merkle envisions very small (submicron) devices floating in a fluid and getting both power and control from pressure impulses. A different approach would be to anchor nanomanipulators on a substrate that could provide power, control, and a transport system for materials. Such a system could move a large number of manipulators by moving the substrate, and thus could build objects much larger than itself in all dimensions.
Simultaneous evolution of mechanochemistry, nanomanipulation, computer-aided design and manufacturing software, and overall system design is the best way to advance each of these individual goals as well as the overall goal. This seems like a lot of simultaneous development, and it is, but trying to do any one of these in isolation from the others would be almost as hard, and would hit a dead end. Only by advancing all at once can we move forward.