As in art, so in nature: from Karel Capek's anthropomorphized machines to purpose built platforms and buckets of stuff, Greg J. Smith explores robotics as transformable systems.
IN 1966 THE PIANIST Cecil Taylor poetically described the creative potential of free jazz in the liner notes to his opus, Unit Structures. “The player advances to the area,” he wrote, “an unknown totality, made whole thru self analysis (improvisation), the conscious manipulation of known material; each piece is choice; architecture…” Unit Structures became a benchmark in unstructured atonal jazz. The album is wild, challenging and nowhere near as modular as the title might imply. During those legendary sessions, Taylor would hammer out percussive chord clusters as cues for his bandmates' improvisations. One can’t help but listen to it without at least partially identifying with his description of musical fragments as building blocks: within the call and response of improvisation "each piece is choice; architecture…"
As in art, so in nature. In a 2002 article in Science magazine, authors George M. Whitesides and Barros Grzybowski identify self-assembly as a “limitlessly elastic” means of describing how “separate or distinct parts of a disordered structure” interact and connect to form new, more complex entities. This short text is an impressive multidisciplinary survey of chemistry, molecular biology, nanotechnology and manufacturing. Its authors outline how emergence occurs at the molecular level (such as in the formation of crystals) all the way up to planetary scale (where the same type of processes drive the formation of weather systems). Whiteside and Grzybowski argue that self-assembly is a ubiquitous phenomenon with numerous applications, and propose a basic framework for categorizing this broad field of inquiry.
"The future of modular robotics is about enabling us to live better lives. For me that means making it easier and more fun to interact with other people. Modular robotics offers the potential of a new type of ‘smart stage’ to create dynamic experiences with others"
One research initiative that would fall under this domain is self-configuring modular robotics, an exploration of robotics as transformable systems rather than fixed objects. From Karel Čapek’s 1921 play R.U.R. (Rossum’s Universal Robots) onwards, the image of anthropomorphic machines has persisted in science fiction, popular culture and consumer electronics. Deployments of robotics within the automotive industry and various international space programs, however, suggest that robots don’t have to look like anything at all, and that it is most efficient to engineer mechanized tools that are fit to purpose, designed to handle the physical requirements of the tasks they are meant to perform.
Modular robotic systems are composed of small, individual building blocks equipped with standardized microcontrollers, sensors, batteries and moving parts for locomotion. The individual units dock with one another and enable the transfer of mechanical forces, moments and power. They form assemblies that are capable of performing rudimentary tasks and rearranging shape.
The M-TRAN3 is a modular robotics project developed by a team at the Institute of Advanced Industrial Science and Technology led by Satoshi Murata. A montage of scenes in an M-TRAN3 demonstration video highlights the impressive flexibility of this latest iteration of the M-TRAN series, which is capable of assembling itself as a quadripedal walker, as a slithering linear structure and as a dense mat of lateral weaves. The object confronts relatively minor changes in topography; the manner in which it reconfigures itself to scale obstacles is uncanny. One might imagine the device maneuvering quite deftly within extremely constrained spaces.
Another well-known modular robotics venture is the Molecube system. Its developers, Hod Lipson and a team at the Cornell Computational Synthesis Lab initially envisioned Molecube as a self-replicating object. The platform has more recently served as the basis for an open-source project aimed at promoting robotics research. A decade ago it would have been inconceivable that the schematics, parts list and related 3D printer-ready files could have been released to the general public. The fact that the project is now out there, ready to be engaged, is a testament to the decreasing cost of digital fabrication, electronic components and microcontrollers. Given the customized chassis geometry of the M-TRAN3, Molecube and other modular robotic systems, the propagation of these projects is fundamentally connected to advances in digital manufacturing.
Robert Miles Kemp is president of Los Angeles based Variate Labs. Kemp co-authored Interactive Architcture, a comprehensive survey of design and emerging technology that delves into the robotic systems discussed in this article. His creative practice deals with the synthesis of architectural, interaction and experience design. When I asked him how research initiatives like Molecube might be deployed in domestic space, he offered that "The future of modular robotics is about enabling us to live better lives. For me that means making it easier and more fun to interact with other people. Modular robotics offers the potential of a new type of ‘smart stage’ to create dynamic experiences with others."
This notion of a "smart stage" is key: using a modular system to build furniture gives it the capacity to accommodate to varying needs. Kemp questions: “why not have a sink that could change size and counters that automatically adjusts to optimize the tasks I am performing?” Window assemblies have been operable for centuries and we’re quite accustomed to adjusting the temperature of our homes and workplaces – are forecasts of reflexive objects and “networked furniture” that far-fetched?
One of the more whimsical endgames discussed within the modular robotics community is the “bucket of stuff”: a container of units that can be used to perform any basic task or function. Perhaps this is the future of manufacturing – on the spot improvisation, a chair that builds itself. While we won’t be seeing an IKEA bucket of stuff anytime soon, experiments with self-configuring modular robotics are currently being conducted in both search and rescue and space exploration contexts. As these kinds of systems begin to increase in scale and flexibility, we will undoubtedly see many more applications in the coming years.