Publications

2019
Melenbrink, Nathan, and Justin Werfel. 2019. “A Swarm Robot Ecosystem For Autonomous Construction, 2017.” Robotic Building: Architecture in the Age of Automation, edited by Gilles Retsin, Manuel Jimenez, Mollie Claypool, and Vicente Soler, 88-90. München: DETAIL.
Carey, Nicole E., Daniel S. Calovi, Paul M. Bardunias, J. Scott Turner, Radhika Nagpal, and Justin Werfel. 2019. “Differential construction response to humidity by related species of mound-building termites.” Journal of Experimental Biology. Abstract
Macrotermes michaelseni and M. natalensis are two morphologically similar species occupying the same habitat across southern Africa. Both build large mounds and tend mutualistic fungal symbionts for nutrients, but despite these behavioural and physiological similarities, the mound superstructures they create differ markedly. The behavioural differences behind this discrepancy remain elusive, and are the subject of ongoing investigations. Here we show that the two species demonstrate distinctive building activity in a lab-controlled environment consisting of still air with low ambient humidity. In these conditions, M. michaelseni transports less soil from a central reservoir, deposits this soil over a smaller area, and creates structures with a smaller volumetric envelope than M. natalensis. In high humidity, no such systematic difference is observed. This result suggests a differential behavioural threshold or sensitivity to airborne moisture that may relate to the distinct macro-scale structures observed in the African bushland.
Melenbrink, Nathan, and Justin Werfel. 2019. “Autonomous Sheet Pile Driving Robots for Soil Stabilization.” 2019 International Conference on Robotics and Automation (ICRA). Montreal, Canada: IEEE. Abstract
Soil stabilization is a fundamental component of nearly all construction projects, ranging from commercial construction to environmental restoration projects. Previous work in autonomous construction has generally not considered these essential stabilization and anchoring tasks. In this work we present Romu, an autonomous robot capable of building continuous linear structures by using a vibratory hammer to drive interlocking sheet piles into soil. We report on hardware parameters and their effects on pile driving performance, and demonstrate autonomous operation in both controlled and natural environments. Finally, we present simulations in which a small swarm of robots build with sheet piles in example terrains, or apply an alternate spray-based stabilizing agent, and quantify the ability of each intervention to mitigate hydraulic erosion.
Calovi, Daniel S., Paul Bardunias, Nicole Carey, J. Scott Turner, Radhika Nagpal, and Justin Werfel. 2019. “Surface curvature guides early construction activity in mound-building termites.” Philosophical Transactions of the Royal Society B 374 (1774): 20180374. Abstract
Termite colonies construct towering, complex mounds, in a classic example of distributed agents coordinating their activity via interaction with a shared environment. The traditional explanation for how this coordination occurs focuses on the idea of a "cement pheromone", a chemical signal left with deposited soil that triggers further deposition. Recent research has called this idea into question, pointing to a more complicated behavioral response to cues perceived with multiple senses. In this work, we explored the role of topological cues in affecting early construction activity in Macrotermes. We created artificial surfaces with a known range of curvatures, coated them with nest soil, placed groups of major workers on them, and evaluated soil displacement as a function of location at the end of one hour. Each point on the surface has a given curvature, inclination, and absolute height; to disambiguate these factors, we conducted experiments with the surface in different orientations. Soil displacement activity is consistently correlated with surface curvature, and not with inclination nor height. Early exploration activity is also correlated with curvature, to a lesser degree. Topographical cues provide a long-term physical memory of building activity in a manner that ephemeral pheromone labeling cannot. Elucidating the roles of these and other cues for group coordination may help provide organizing principles for swarm robotics and other artificial systems.
2018
Gershenson, Carlos, Vito Trianni, Justin Werfel, and Hiroki Sayama. 2018. “Self-Organization and Artificial Life: A Review.” The 2018 International Conference on Artificial Life (ALIFE 2018). Abstract
Self-organization has been an important concept within a number of disciplines, which Artificial Life (ALife) also has heavily utilized since its inception.  The term and its implications, however, are often confusing or misinterpreted.  In this work, we provide a mini-review of self-organization and its relationship with ALife, aiming at initiating discussions on this important topic with the interested audience. We first articulate some fundamental aspects of self-organization, outline its usage, and review its applications to ALife within its soft, hard, and wet domains. We also provide perspectives for further research.
Melenbrink, Nathan, and Justin Werfel. 2018. “Local force cues for strength and stability in a distributed robotic construction system.” Swarm Intelligence 12 (2): 129-153. Abstract
Construction of spatially extended, self-supporting structures requires a consideration of structural stability throughout the building sequence. For collective construction systems, where independent agents act with variable order and timing under decentralized control, ensuring stability is a particularly pronounced challenge. Previous research in this area has largely neglected considering stability during the building process. Physical forces present throughout a structure may be usable as a cue to inform agent actions as well as an indirect communication mechanism (stigmergy) to coordinate their behavior, as adding material leads to redistribution of forces which then informs the addition of further material. Here we consider in simulation a system of decentralized climbing robots capable of traversing and extending a two-dimensional truss structure, and explore the use of feedback based on force sensing as a way for the swarm to anticipate and prevent structural failures. We consider a scenario in which robots are tasked with building an unsupported cantilever across a gap, as for a bridge, where the goal is for the swarm to build any stable spanning structure rather than to construct a specific predetermined blueprint. We show that access to local force measurements enables robots to build cantilevers that span significantly farther than those built by robots without access to such information. This improvement is achieved by taking measures to maintain both strength and stability, where strength is ensured by paying attention to forces during locomotion to prevent joints from breaking, and stability is maintained by looking at how loads transfer to the ground to ensure against toppling. We show that swarms that take both kinds of forces into account have improved building performance, in both structured settings with flat ground and unpredictable environments with rough terrain.
2017
Melenbrink, Nathan, Paul Kassabian, Achim Menges, and Justin Werfel. 2017. “Towards Force-aware Robot Collectives for On-site Construction.” Association for Computer Aided Design in Architecture (ACADIA), 382-391. Abstract
Due to the irregular and variable environments in which most construction projects take place, the topic of on-site automation has previously been largely neglected in favor of off-site prefabrication. While prefabrication has certain obvious economic and schedule benefits, a number of potential applications would benefit from a fully autonomous robotic construction system capable of building without human supervision or intervention -- for example, building in remote environments, or building structures whose form changes over time. Previous work using a swarm approach to robotic assembly generally neglected to consider forces acting on the structure, which is necessary to guarantee against failure during construction. In this paper we report on key findings for how distributed climbing robots can use local force measurements to assess aspects of global structural state. We then chart out a broader trajectory for the affordances of distributed on-site construction in the built environment and position our contributions within this research agenda. The principles explored in simulation are demonstrated in hardware, including solutions for force-sensing as well as a climbing robot.
Melenbrink, Nathan, Panagiotis Michalatos, Paul Kassabian, and Justin Werfel. 2017. “Using Local Force Measurements to Guide Construction by Distributed Climbing Robots.” IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Abstract
Construction automation has historically been driven by top-down implementations of specific tasks, which are neither responsive nor resilient to dynamic situations, and often require centralized control or human supervision. Previous work on robotic assembly has generally neglected to consider forces acting on the structure, whether in the completed structure alone or throughout the building process. In this paper, we investigate the utility of local force measurements in guiding construction by a distributed team of strut-climbing robots, focusing on a scenario involving building an unsupported span out across a gap in a two-dimensional vertical plane, as a step towards building a bridge. We show that such measurements enable robots to build structures that cantilever significantly further than those built by robots without access to such information, while maintaining stability throughout the building sequence. We consider both structures securely anchored to the ground and those resting unanchored atop it, using a counterbalancing approach in the latter case to permit cantilevering. The principles explored in simulation are also demonstrated in hardware, including a prototype strut-climbing robot and truss components, incorporating a cost-effective sensor implementation that reports the requisite force information.
Green, Ben, Paul Bardunias, J. Scott Turner, Radhika Nagpal, and Justin Werfel. 2017. “Excavation and aggregation as organizing factors in de novo construction by mound-building termites.” Proceedings of the Royal Society B 284 (1856): 20162730. Abstract

Termites construct complex mounds that are orders of magnitude larger than any individual and fulfill a variety of functional roles. Yet the processes through which these mounds are built, and by which the insects organize their efforts, remain poorly understood. The traditional understanding focuses on stigmergy, a form of indirect communication in which actions that change the environment provide cues that influence future work. Termite construction has long been thought to be organized via a putative "cement pheromone": a chemical added to deposited soil that stimulates further deposition in the same area, thus creating a positive feedback loop whereby coherent structures are built up. To investigate the detailed mechanisms and behaviors through which termites self-organize the early stages of mound construction, we tracked the motion and behavior of major workers from two Macrotermes species in experimental arenas. Rather than a construction process focused on accumulation of depositions, as models based on cement pheromone would suggest, our results indicated that the primary organizing mechanisms were based on excavation. Digging activity was focused on a small number of excavation sites, which in turn provided templates for soil deposition. This behavior was mediated by a mechanism of aggregation, with termites being more likely to join in the work at an excavation site as the number of termites presently working at that site increased. Statistical analyses showed that this aggregation mechanism was a response to active digging, distinct from and unrelated to putative chemical cues that stimulate deposition. Agent-based simulations quantitatively supported the interpretation that the early stage of de novo construction is primarily organized by excavation and aggregation activity rather than by stigmergic deposition.

Werfel, Justin, Donald E. Ingber, and Yaneer Bar-Yam. 2017. “Theory and associated phenomenology for intrinsic mortality arising from natural selection.” PLOS ONE 12 (3): e0173677. Publisher's Version Abstract

Standard evolutionary theories of aging and mortality, implicitly based on assumptions of spatial averaging, hold that natural selection cannot favor shorter lifespan without direct compensating benefit to individual reproductive success. However, a number of empirical observations appear as exceptions to or are difficult to reconcile with this view, suggesting explicit lifespan control or programmed death mechanisms inconsistent with the classic understanding. Moreover, evolutionary models that take into account the spatial distributions of populations have been shown to exhibit a variety of self-limiting behaviors, maintained through environmental feedback. Here we extend recent work on spatial modeling of lifespan evolution, showing that both theory and phenomenology are consistent with programmed death. Spatial models show that self-limited lifespan robustly results in long-term benefit to a lineage; longer-lived variants may have a reproductive advantage for many generations, but shorter lifespan ultimately confers long-term reproductive advantage through environmental feedback acting on much longer time scales. Numerous model variations produce the same qualitative result, demonstrating insensitivity to detailed assumptions; the key conditions under which self-limited lifespan is favored are spatial extent and locally exhaustible resources. Factors including lower resource availability, higher consumption, and lower dispersal range are associated with evolution of shorter lifespan. A variety of empirical observations can parsimoniously be explained in terms of long-term selective advantage for intrinsic mortality. Classically anomalous empirical data on natural lifespans and intrinsic mortality, including observations of longer lifespan associated with increased predation, and evidence of programmed death in both unicellular and multicellular organisms, are consistent with specific model predictions. The generic nature of the spatial model conditions under which intrinsic mortality is favored suggests a firm theoretical basis for the idea that evolution can quite generally select for shorter lifespan directly.

Carey, Nicole, Radhika Nagpal, and Justin Werfel. 2017. “Fast, accurate, small-scale 3D scene capture using a low-cost depth sensor.” WACV 2017. Abstract

Commercially available depth sensing devices are primarily designed for domains that are either macroscopic, or static.  We develop a solution for fast microscale 3D reconstruction, using off-the-shelf components.  By the addition of lenses, precise calibration of camera internals and positioning, and development of bespoke software, we turn an infrared depth sensor designed for human-scale motion and object detection into a device with mm-level accuracy capable of recording at up to 30Hz.

2015
Werfel, Justin, Donald E. Ingber, and Yaneer Bar-Yam. 2015. “Programmed Death is Favored by Natural Selection in Spatial Systems.” Physical Review Letters 114: 238103. Publisher's Version Abstract

Standard evolutionary theories of aging and mortality, implicitly based on mean-field assumptions, hold that programed mortality is untenable, as it opposes direct individual benefit. We show that in spatial models with local reproduction, programed deaths instead robustly result in long-term benefit to a lineage, by reducing local environmental resource depletion via spatiotemporal patterns causing feedback over many generations. Results are robust to model variations, implying that direct selection for shorter life span may be quite widespread in nature.

Rubenstein, Michael, Bo Cimino, Radhika Nagpal, and Justin Werfel. 2015. “AERobot: An Affordable One-Robot-Per-Student System for Early Robotics Education.” IEEE International Conference on Robotics and Automation (ICRA). PDF Abstract

There is a widely recognized need for improved STEM education and increased technological literacy. Robots represent a promising educational tool with potentially large impact, due to their broad appeal and wide relevance; however, many existing educational robot platforms have cost as a barrier to widespread use. Here we present AERobot, a simple low-cost robot that can be easily used for introductory programming and robotics teaching, starting from a primary or middle school level. The hardware is open-source and can be built for ~$10 per robot, making it possible for each student to have (and keep) their own robot, while still encompassing a rich sensor suite enabling a variety of activities. A free, open-source graphical programming environment allows students without previous programming experience to command the robot. We report on the results of three sessions of a one-week pilot course held in the summer of 2014 by STEM summer camp i2 Camp.

Grun, Casey, Justin Werfel, David Yu Zhang, and Peng Yin. 2015. “DyNAMiC Workbench: an integrated development environment for dynamic DNA nanotechnology.” Journal of the Royal Society Interface 12 (111). Publisher's Version Abstract

Dynamic DNA nanotechnology provides a promising avenue for implementing sophisticated assembly processes, mechanical behaviours, sensing and computation at the nanoscale. However, design of these systems is complex and error-prone, because the need to control the kinetic pathway of a system greatly increases the number of design constraints and possible failure modes for the system. Previous tools have automated some parts of the design workflow, but an integrated solution is lacking. Here, we present software implementing a three ‘tier’ design process: a high-level visual programming language is used to describe systems, a molecular compiler builds a DNA implementation and nucleotide sequences are generated and optimized. Additionally, our software includes tools for analysing and ‘debugging’ the designs in silico, and for importing/exporting designs to other commonly used software systems. The software we present is built on many existing pieces of software, but is integrated into a single package—accessible using a Web-based interface at http://molecular-systems.net/workbench. We hope that the deep integration between tools and the flexibility of this design process will lead to better experimental results, fewer experimental design iterations and the development of more complex DNA nanosystems.

Petersen, Kirstin, Paul Bardunias, Nils Napp, Justin Werfel, Radhika Nagpal, and J. Scott Turner. 2015. “Arrestant property of recently manipulated soil on Macrotermes michaelseni as determined through visual tracking and automatic labeling of individual termite behaviors.” Behavioural Processes 116: 8-11. Publisher's Version Abstract

The construction of termite nests has been suggested to be organized by a stigmergic process that makes use of putative cement pheromone found in saliva and recently manipulated soil (“nest material”), hypothesized to specifically induce material deposition by workers. Herein, we tracked 100 individuals placed in arenas filled with a substrate of half nest material, half clean soil, and used automatic labeling software to identify behavioral states. Our findings suggest that nest material acts to arrest termites; termites prefer to spend time on nest material when compared against clean soil. Residency time was significantly greater, and all construction behaviors occurred significantly more often on nest material. The arrestant function of nest material must be accounted for in experiments that seek semiochemical cues for the organization of labor. Future research will focus on the manner in which termites combine olfaction with tactile cues as well as other organizing factors during construction.

2014
Designing Collective Behavior in a Termite-Inspired Robot Construction Team
Werfel, Justin, Kirstin Petersen, and Radhika Nagpal. 2014. “Designing Collective Behavior in a Termite-Inspired Robot Construction Team.” Science 343 (6172): 754-758. Publisher's Version Abstract

Complex systems are characterized by many independent components whose low-level actions produce collective high-level results. Predicting high-level results given low-level rules is a key open challenge; the inverse problem, finding low-level rules that give specific outcomes, is in general still less understood. We present a multi-agent construction system inspired by mound-building termites, solving such an inverse problem. A user specifies a desired structure, and the system automatically generates low-level rules for independent climbing robots that guarantee production of that structure. Robots use only local sensing and coordinate their activity via the shared environment. We demonstrate the approach via a physical realization with three autonomous climbing robots limited to onboard sensing. This work advances the aim of engineering complex systems that achieve specific human-designed goals.

For free access without a subscription, please follow the links here.

Werfel, Justin. 2014. “Embodied Teachable Agents: Learning by Teaching Robots.” New Research Frontiers for Intelligent Autonomous Systems (NRF-IAS-2014). PDF Abstract

Robots have an untapped potential to contribute to education by acting as subordinate learners for students to teach. The benefits that the act of teaching provides for one’s own learning have long been recognized; tutoring-associated improvements in measures like achievement scores, depth of understanding, and motivation are often far greater for the tutor than the tutee. Artificial agents can help students reap these benefits by providing surrogate pupils for them to teach, with potential advantages over human tutees. Robots have been observed to be more effective and compelling than virtual agents in a variety of contexts. However, research on teachable agents for education has been limited to virtual agents, while research on humans teaching robots has been concerned with learning for the benefit of the robot rather than that of the human. A new research direction exploring robots as teachable agents will lead to widespread benefits in education, and open new possibilities enabled by physical embodiment.

Rutherford, Alex, Dion Harmon, Justin Werfel, Alexander S. Gard-Murray, Shlomiya Bar-Yam, Andreas Gros, Ramon Xulvi-Brunet, and Yaneer Bar-Yam. 2014. “Good Fences: The Importance of Setting Boundaries for Peaceful Coexistence.” PLOS ONE 9 (5): e95660. Publisher's Version Abstract

We consider the conditions of peace and violence among ethnic groups, testing a theory designed to predict the locations of violence and interventions that can promote peace. Characterizing the model's success in predicting peace requires examples where peace prevails despite diversity. Switzerland is recognized as a country of peace, stability and prosperity. This is surprising because of its linguistic and religious diversity that in other parts of the world lead to conflict and violence. Here we analyze how peaceful stability is maintained. Our analysis shows that peace does not depend on integrated coexistence, but rather on well defined topographical and political boundaries separating groups, allowing for partial autonomy within a single country. In Switzerland, mountains and lakes are an important part of the boundaries between sharply defined linguistic areas. Political canton and circle (sub-canton) boundaries often separate religious groups. Where such boundaries do not appear to be sufficient, we find that specific aspects of the population distribution guarantee either sufficient separation or sufficient mixing to inhibit intergroup violence according to the quantitative theory of conflict. In exactly one region, a porous mountain range does not adequately separate linguistic groups and that region has experienced significant violent conflict, leading to the recent creation of the canton of Jura. Our analysis supports the hypothesis that violence between groups can be inhibited by physical and political boundaries. A similar analysis of the area of the former Yugoslavia shows that during widespread ethnic violence existing political boundaries did not coincide with the boundaries of distinct groups, but peace prevailed in specific areas where they did coincide. The success of peace in Switzerland may serve as a model to resolve conflict in other ethnically diverse countries and regions of the world.