Kinematic Self-Replicating Machines

© 2004 Robert A. Freitas Jr. and Ralph C. Merkle. All Rights Reserved.

Robert A. Freitas Jr., Ralph C. Merkle, Kinematic Self-Replicating Machines, Landes Bioscience, Georgetown, TX, 2004.


3.17 Lohn Electromechanical Replicators (1998)

Following in the spirit of the simple early devices of Penrose (Section 3.3) and Morowitz (Section 3.5), in 1998 Lohn et al [1157] presented designs for two simple electromechanical replicators that could be constructed out of plastic, batteries and electromagnets. Following the invariant design principles exemplified by earlier devices, Lohn’s systems draw analogy from chemistry, with electromagnetic forces modeling molecular bonding, complementary physical shapes modeling molecular structure, and electrical current flow modeling activation. Lohn notes that “designing a self-assembling structure is akin to engineering an artificial catalyst, where a specific event becomes likely only in the presence of specific components. This problem appears simple but is deceptive: it is not very difficult to engineer the self-assembly process where a parent structure attracts the necessary components, assembles them, then detaches from the offspring structure. The difficulty lies in ensuring robustness. One must prevent the possibility of blocked active sites, cancerous growths, crystal-growth formations, and other deleterious side effects that may occur when components are randomly interacting.”

Lohn follows Penrose’s ground rules [681] which define a “seed” structure as self-replicating if it can induce the formation of two or more new structures which are identical to itself and which are assembled by combining simpler components already present in the environment: “Assume we have a two-dimensional environment in which components are mobile (e.g., coins placed between two parallel plates of glass). If we populate the environment with numerous components, add energy to it (e.g., by shaking), and wait, we expect the initial and final configurations of the system to be similar – randomly placed components. However, if we place a self-assembling seed structure into the initial environment, then add energy, we would expect to see multiple seed copies appear over time, with a corresponding decrease in the number of free components.”

Lohn’s two electromechanical models – the first a seed composed of two identical components, the second a seed composed of two distinct components – are inspired by Morowitz’s 1959 model [688] (Section 3.5) while making the design simpler using only active components. Lohn’s designs have no mechanical linkages between components and consist solely of circuits, whereas the Morowitz model relies on two sliding parts to operate two switches. All of Lohn’s components are active, whereas Morowitz’ model used one active and one passive component. Finally, Lohn’s components have no explicit switches – circuits are switched on and off via the bonds that form between components – and his designs require one battery rather than two. Both devices are intended to replicate when placed in a “sea” of parts jostling on an agitated planar surface.

Lohn Monotype Electromechanical Replicator. The monotype replicating seed consists of two identical components, each consisting of a pair of electromagnets cross-wired through a set of contacts, a resistor, and a battery (Figure 3.55). All parts adhere solely by magnetic attraction. Replication proceeds as follows (Figure 3.56): (I) The monotype seed is formed by forcing two elements (each with only one exposed south terminal) together. (II) Forming the seed closes contacts that flip the polarity of the rightmost electromagnet and energizes the second electromagnet in the same (left) element, locking the two “seed” elements. The south terminal of a single element is attracted to the newly energized north “seed” terminal. (III) The new threesome closes circuits that strengthen the existing charges and energize the leftmost electromagnet, exposing a north terminal. As in the previous step, a single element is attracted to the exposed north terminal. (IV) The attachment of the fourth element closes a circuit that flips the sign of the middle-left electromagnet (parallel to the original seed formation) causing the arrangement to become unstable. The two north terminals in the center of the foursome now repel each other. (V) The final result is a new 2-element seed in addition to the first 2-element seed. Over time the single elements with an exposed south terminal are transformed into 2-element “seeds” with a north terminal exposed.

Lohn Polytype Electromechanical Replicator. The polytype replicating seed consists of two distinct components (Figure 3.57). The first component is a pair of electromagnets wired to a set of contacts. The second component is a single electromagnet and a battery wired to a different set of contacts. Again, all parts adhere solely by magnetic attraction. Replication proceeds as follows (Figure 3.58): (I) The polytype seed forms via random collision or by artificial introduction; circuit 1 is energized forming a bond between components A and B. (II) When immersed in a sea of individual components, only component B is biased towards attaching to the AB complex; an attractive force and complementary shapes encourage B to attach. (III) Likewise, the AB-B complex encourages component A to attach as shown. (IV) Once attached, the lower AB complex is energized, and the resulting repellant force pushes the offspring complex away from its parent.

Lohn and his colleagues attempted to build both models from scratch, winding the specialized electromagnets by hand, but were unable to complete these constructions due to serious limitations in the time and resources available [1158]. The authors of this book can see no reason why Lohn’s models, if physically implemented, would not work. Lohn et al [1157] conclude: “The electromechanical models presented here represent one medium in which self-assembling machines can be studied. The future of designing such models for nanotechnology will likely benefit from work in many disciplines, ranging from biochemistry to physics to artificial life. Because designing self-assembling systems is non-trivial, automating the design process [371] may hold great benefit. Furthermore, automated approaches may create self-assembling systems that embody principles human designers would never think of.”


Last updated on 1 August 2005