Warren Christopher Ruder

Programmable materials integrated with intelligent signal processing are essential for developing adaptive and autonomous environmental sensing systems. Here, we present a biohybrid material platform in which environmental DNA is captured by streptavidin (SA)-functionalized microparticles and delivered to a cell-free transcription-translation system for intelligent information processing. The microparticles selectively capture biotinylated DNA from their surroundings, which serve as molecular algorithm inputs for downstream biological computation. DNA immobilized on the particle surface remains transcriptionally active, enabling cell-free systems to produce tunable outputs, including (1) linear responses proportional to DNA input concentration and (2) threshold-dependent switching behaviors achieved through competitive binding. By offloading computation to a cell-free biochemical processor, this approach enables scalable, biocompatible sensing without the need for onboard electronics. This biohybrid design provides a versatile framework for engineering intelligent materials with applications in developing smart tools for healthcare and environmental monitoring.

Developing resilience against climate change and establishing food security will require significant research into the responses of multicellular organisms to their environment. New approaches, such as lab automation, can substantially increase the rate of data collection for organism‐level behavior. This report describes an automated robotic system for studying multicellular organisms to accelerate scientific experimentation. The robot can simultaneously image and deliver chemicals to each organism during multiday experiments, and uses a deep learning model to automatically obtain phenotypic data. This system's abilities are demonstrated by creating a plant growth strategy for food security applications that increases biomass while decreasing nutrient utilization. Furthermore, plant growth is characterized under high salt concentrations to better understand the effects of climate change on freshwater ecosystems. This robotic approach improves lab automation for studying multicellular organisms by increasing experimental throughput, and will enable researchers to improve crop yields under uncertain climates and predict the response of organisms in changing environments.

The communication between different cell populations is an important aspect of many natural phenomena that can be studied with microfluidics. Using microfluidic valves, these complex interactions can be studied with a higher level of control by placing a valve between physically separated populations. However, most current valve designs do not display the properties necessary for this type of system, such as providing variable flow rate when embedded inside a microfluidic device. While some valves have been shown to have such tunable behavior, they have not been used for dynamic, real-time outputs. We present an electric solenoid valve that can be fabricated completely outside of a cleanroom and placed into any microfluidic device to offer control of dynamic fluid flow rates and profiles. After characterizing the behavior of this valve under controlled test conditions, we developed a regression model to determine the required input electrical signal to provide the solenoid the ability to create a desired flow profile. With this model, we demonstrated that the valve could be controlled to replicate a desired, time-varying pattern for the interface position of a co-laminar fluid stream. Our approach can be performed by other investigators with their microfluidic devices to produce predictable, dynamic fluidic behavior. In addition to modulating fluid flows, this work will be impactful for controlling cellular communication between distinct populations or even chemical reactions occurring in microfluidic channels.

Swimming microrobots are increasingly developed with complex materials and dynamic shapes and are expected to operate in complex environments in which the system dynamics are difficult to model and positional control of the microrobot is not straightforward to achieve. Deep reinforcement learning is a promising method of autonomously developing robust controllers for creating smart microrobots, which can adapt their behavior to operate in uncharacterized environments without the need to model the system dynamics. This article reports the development of a smart helical magnetic hydrogel microrobot that uses the soft actor critic reinforcement learning algorithm to autonomously derive a control policy which allows the microrobot to swim through an uncharacterized biomimetic fluidic environment under control of a time‐varying magnetic field generated from a three‐axis array of electromagnets. The reinforcement learning agent learns successful control policies from both state vector input and raw images, and the control policies learned by the agent recapitulate the behavior of rationally designed controllers based on physical models of helical swimming microrobots. Deep reinforcement learning applied to microrobot control is likely to significantly expand the capabilities of the next generation of microrobots.

Swimming microrobots are increasingly developed with complex materials and dynamic shapes and are expected to operate in complex environments in which the system dynamics are difficult to model and positional control of the microrobot is not straightforward to achieve. Deep reinforcement learning is a promising method of autonomously developing robust controllers for creating smart microrobots, which can adapt their behavior to operate in uncharacterized environments without the need to model the system dynamics. Here, we report the development of a smart helical magnetic hydrogel microrobot that used the soft actor critic reinforcement learning algorithm to autonomously derive a control policy which allowed the microrobot to swim through an uncharacterized biomimetic fluidic environment under control of a time varying magnetic field generated from a three-axis array of electromagnets. The reinforcement learning agent learned successful control policies with fewer than 100,000 training steps, demonstrating sample efficiency for fast learning. We also demonstrate that we can fine tune the control policies learned by the reinforcement learning agent by fitting mathematical functions to the learned policy's action distribution via regression. Deep reinforcement learning applied to microrobot control is likely to significantly expand the capabilities of the next generation of microrobots.

An evolving understanding of disease pathogenesis has compelled the development of new drug delivery approaches. Recently, bioinspired microrobots have gained traction as drug delivery systems. By leveraging the microscale phenomena found in physiological systems, these microrobots can be designed with greater maneuverability, which enables more precise, controlled drug release. Their function could be further improved by testing their efficacy in physiologically relevant model systems as part of their development. In parallel with the emergence of microscale robots, organ-on-a-chip technologies have become important in drug discovery and physiological modeling. These systems reproduce organ-level functions in microfluidic devices, and can also incorporate specific biological, chemical, and physical aspects of a disease. This review highlights recent developments in both microrobotics and organ-on-a-chip technologies and envisions their combined use for developing future drug delivery systems.

The integration of molecular robots and synthetic biology allows for the creation of sophisticated behaviors at the molecular level. Similar to the synergy between bioelectronics and soft robotics, synthetic biology provides control circuitry for molecular robots. By encoding perception-action modules within synthetic circuits, molecular machines can advance beyond repeating tasks to the incorporation of complex behaviors. In particular, cell-free synthetic biology provides biomolecular circuitry independent of living cells. This research update reviews the current progress in using synthetic biology as perception-action control modules in robots from molecular robots to macroscale robots. Additionally, it highlights recent developments in molecular robotics and cell-free synthetic biology and suggests their combined use as a necessity for future molecular robot development.

Microfluidic technologies are frequently employed as point-of-care diagnostic tools for improving time-to-diagnosis and improving patient outcomes in clinical settings. These microfluidic devices often are designed to operate with peripheral equipment for liquid handling that increases the cost and complexity of these systems and reduces their potential for widespread adoption in low resource healthcare applications. Here, we present a low-cost (~$120), open-source peristaltic pump constructed with a combination of three dimensional (3D)-printed parts and common hardware, which is amenable to deployment with microfluidic devices for point-of-care diagnostics. This pump accepts commonly available silicone rubber tubing in a range of sizes from 1.5 to 3 mm, and is capable of producing flow rates up to 1.6 mL min⁻¹. This device is programmed with an Arduino microcontroller, allowing for custom flow profiles to fit a wide range of low volume liquid handling applications including precision liquid aliquoting, flow control within microfluidics, and generation of physiologically relevant forces for studying cellular mechanobiology within microfluidic systems.