Research

Collective Construction

Many social insects build functional structures collectively, sometimes from their own bodies and sometimes from inanimate materials. These structures serve essential roles, including shelter, thermoregulation, and even capturing prey. What makes this especially remarkable is that the structures they build are often formed in very noisy and messy environments, yet they exhibit mechanical stability, structural integrity, and adaptability to their surroundings.

Unlike human-engineered designs, these structures are built without centralized control, technical drawings, or even a pre-decided path plan. Rather, what emerges from the combined efforts of many collaborating individual agents is the result of decentralized self-organization. Each agent follows simple rules for depositing or removing material based on local information sensed from its immediate surroundings. The sum of these behaviors result in the remarkable structures we observe in nature.

We investigate the principles behind collective construction observed in tent caterpillars, fire ants, and bees, among other notable examples, with the goal of translating this knowledge into novel modes of engineering design. We believe this work has great potential to impact how autonomous robots attempt manufacturing in the most challenging environments, where redundancy is paramount and remote control or centralized computation is infeasible, such as deep within our oceans and caves or even distant planets.

 Active Matter within Soft Boundaries

Active matter is a class of materials whose constituents consume energy to generate internal forces or stresses, driving the system out of equilibrium and giving rise to self-driven motion. At the level of individual particles, self-propelled active matter exhibits persistent random motion, meaning that particles move along relatively straight trajectories and undergo random reorientations due to internal and environmental noise.

When self-propelled active particles are confined within soft or deformable boundaries, their modes of motion can change dramatically, depending on how the particles interact with the boundary. These interactions are influenced by the boundary’s material properties, mass, and geometry, as well as by the nature of the particle–boundary coupling. We study how these parameters shape emergent locomotion in the coupled particle–boundary systems using simple active agents (small bristle bots) confined within elastic boundaries.

These experiments provide a controlled platform for understanding how coupling between active matter and soft materials influences the behavior of both, and how locomotion can be directed through geometry, compliance, and interaction rules rather than explicit control.

Directed Nucleation and Growth

In many natural and engineered systems, structures grow from multiple nucleation sites rather than a single origin. As growth fronts expand and interact, their placement, relative growth rates, and boundary geometry determine how uniformly a target region is covered. This principle underlies phenomena ranging from crystal formation and biological growth to viral spreading, illustrating how global structure emerges from local initiation and growth rules.

These ideas are especially powerful in frontal polymerization, where growth initiated from a single point produces a spherical front, and multiple initiation sites allow fronts to grow, merge, and form complex geometries. This raises a natural design question: how can one grow asymmetric or irregular shapes using growth fronts that are locally spherical? We investigate this by using ideas from the physics of jamming to determine where nucleation points should be placed in order to produce desired shapes. This approach provides a new route to shape control in growth-based fabrication, with direct applications in growth-based 3D printing.

LeWitt Robots

Imagine a group of simple robots, each equipped with a pen, a basic sensor, and a small microcontroller. Rather than following a global plan, each robot draws lines according to a local rule based on simple measurements (such as distances or angles between the existing lines on the surface). As the robots move and draw, they respond to changes in their environment created by the actions of other robots. Through these local interactions, various shapes and patterns emerge at the collective level. Our goal is to design these local rules so that a desired global pattern reliably emerges, without any centralized control. 

This project bridges our theoretical work on collective construction with ideas from algorithmic art. Inspired in part by artists such as Sol LeWitt, whose work emphasizes rule-based generation over authorship of form, we refer to these systems as LeWitt robots. Using this platform, we investigate how simple robotic agents can collectively produce complex, adaptive structures, offering insight into decentralized robotics, autonomous fabrication, and the broader principles of directed emergence.