Each of these minuscule machines measures approximately 200 by 300 by 50 micrometers, rendering them significantly smaller than a common grain of salt. This diminutive size places them on par with many living microorganisms, opening up unprecedented possibilities for applications such as in-depth cellular monitoring by doctors or the intricate assembly of microscopic components in advanced manufacturing processes. The robots derive their power entirely from light, and crucially, they are equipped with microscopic computers that enable them to follow pre-programmed trajectories, detect subtle variations in local temperature, and dynamically adjust their movements in response to these environmental cues.

This groundbreaking work, detailed in prestigious scientific journals Science Robotics and Proceedings of the National Academy of Sciences (PNAS), distinguishes itself from prior microscopic machines by completely eliminating the need for external power sources like wires, magnetic fields, or remote controls. This liberation from external dependencies makes them the first truly autonomous and programmable robots operating at such an infinitesimally small scale. "We’ve created autonomous robots that are 10,000 times smaller than previously possible," states Marc Miskin, an Assistant Professor in Electrical and Systems Engineering at Penn Engineering and the senior author of the published papers. "This achievement unlocks an entirely new dimension for the development of programmable robots."

The journey to shrinking robots to this unprecedented scale has been fraught with considerable challenges. While electronic components have seen relentless miniaturization over the decades, the field of robotics has lagged significantly behind. According to Professor Miskin, achieving genuine independence in machines smaller than one millimeter has remained an enduring enigma. "Developing robots that can operate autonomously at sizes below one millimeter is an extraordinarily difficult feat," he explains. "The scientific community has essentially been grappling with this problem for the past 40 years."

At everyday scales, the physics governing motion are dominated by forces such as gravity and inertia, which are directly proportional to an object’s volume. However, at the microscopic realm, surface-related forces take precedence. Phenomena like drag and viscosity become overwhelmingly influential, fundamentally altering how movement operates. "When you reach a certain level of smallness, interacting with water becomes akin to pushing through thick tar," Miskin elaborates, illustrating the dramatic shift in physical principles.

The limitations imposed by these altered physics render conventional robotic designs ineffective. Tiny appendages like arms or legs, if conceived, would be exceptionally fragile and prohibitively difficult to manufacture at such minuscule dimensions. "Legs and arms that are extremely small are prone to breaking easily," Miskin points out. "Furthermore, their fabrication presents immense manufacturing hurdles."

To surmount these inherent obstacles, the research team devised an entirely novel approach to robotic locomotion, one that harmonizes with the physics of the microscopic world rather than attempting to overcome them.

The innovative method by which these microscopic robots achieve propulsion is a departure from the strategies employed by larger aquatic organisms. While fish and other macroscopic swimmers propel themselves by pushing water backward, leveraging Newton’s Third Law for forward motion, these tiny robots adopt a fundamentally different tactic.

Instead of relying on physical flexing or bending mechanisms, the robots generate an electrical field. This field gently nudges charged particles within the surrounding liquid. As these ions are propelled, they, in turn, drag adjacent water molecules along with them, effectively inducing fluid motion around the robot. "It’s as if the robot is situated within a moving river," Miskin explains, "except the robot itself is the agent causing that river to flow."

By precisely modulating this electrical field, the robots gain the ability to alter their direction, navigate intricate pathways, and even synchronize their movements in collective formations that bear a resemblance to schools of fish. Their propulsion system allows them to achieve speeds of up to one body length per second.

A significant advantage of this light-activated swimming mechanism is its inherent durability. Because it utilizes electrodes with no moving parts, the robots exhibit remarkable resilience. Professor Miskin notes that they can be repeatedly transferred between different liquid samples using a micropipette without sustaining any damage. Powered by light emitted from an LED, these robots are capable of sustained locomotion for periods extending up to several months.

The challenge of imbuing such a microscopic entity with true autonomy extends beyond mere movement. A truly autonomous robot must possess the capacity to sense its environment, process information to make decisions, and independently manage its power supply. The integration of all these essential components onto a chip measuring only a fraction of a millimeter across presented a formidable engineering hurdle, which was tackled by David Blaauw’s team at the University of Michigan.

Blaauw’s laboratory already held a distinguished record for the creation of the world’s smallest computer. Their collaboration with Miskin’s team was ignited by a serendipitous encounter at a Defense Advanced Research Projects Agency (DARPA) presentation five years prior. They quickly recognized the synergistic potential of their respective technologies. "We immediately saw that Penn Engineering’s propulsion system and our miniature electronic computers were perfectly complementary," recalls Blaauw. Despite this initial synergy, the realization of a functional robot from this concept required an intensive five-year development period.

One of the most significant obstacles encountered was the issue of power management. "The primary challenge for the electronics was the extremely limited power generated by the minuscule solar panels, which output only 75 nanowatts," states Blaauw. "This is over 100,000 times less power than what a typical smartwatch consumes." To overcome this deficit, the team engineered specialized circuits meticulously designed to operate at exceptionally low voltages, thereby reducing power consumption by an order of magnitude exceeding 1,000 times.

Physical space represented another critical constraint. The solar panels occupied the majority of the robot’s surface area, leaving minimal room for essential computing hardware. The researchers addressed this spatial limitation by fundamentally re-architecting the robot’s software. "We had to completely rethink the computer program instructions," Blaauw explains. "This involved condensing what would conventionally require numerous instructions for propulsion control into a single, specialized instruction, thereby shrinking the program’s length to fit within the robot’s severely limited memory capacity."

The culmination of these advancements has resulted in what the researchers believe to be the first sub-millimeter robot capable of genuine decision-making. To their knowledge, no previous effort has succeeded in integrating a complete computer system, comprising a processor, memory, and sensors, into a robot of such diminutive dimensions. This remarkable achievement empowers the robots to perceive their surroundings and react independently to environmental stimuli.

These microscopic robots are equipped with electronic temperature sensors that can discern temperature variations as small as one-third of a degree Celsius. This fine-grained sensing capability enables them to navigate towards warmer regions or to report temperature readings that can serve as indicators of cellular metabolic activity, thus offering a novel method for monitoring individual cells.

The mechanism for communicating these measured temperature values necessitated an ingenious solution. "To transmit their temperature measurements, we devised a special computer instruction that encodes a specific value, such as the measured temperature, into the subtle variations of a short, repetitive movement the robot performs," describes Blaauw. "We then observe this ‘dance’ through a microscope equipped with a camera and decode from these movements what the robots are communicating to us. This process bears a striking resemblance to how honey bees convey information to each other."

The same light source that powers the robots is also employed for their programming. Each robot is assigned a unique address, allowing researchers to upload distinct sets of instructions to individual units. "This capability opens up a vast array of possibilities," Blaauw adds, "with each robot potentially assigned a unique role within a larger, collaborative task."

The current generation of robots represents merely the foundational stage of development. Future iterations are envisioned to incorporate more sophisticated programming, achieve enhanced speeds, integrate additional sensor modalities, and operate effectively in more challenging environments. The researchers deliberately designed the system as a versatile platform, seamlessly combining a robust propulsion method with electronics that are both cost-effective to manufacture and adaptable for future enhancements.

"This is truly just the beginning," emphasizes Miskin. "We have demonstrated that it is possible to embed a ‘brain,’ a sensor, and a motor into something so small it’s almost imperceptible, and have it function reliably for months. Once this fundamental framework is established, it becomes possible to layer on a wide spectrum of intelligence and functionality. This breakthrough paves the way for an entirely new era of robotics at the microscale."

The research was a collaborative effort involving the University of Pennsylvania’s School of Engineering and Applied Science, the Penn School of Arts & Sciences, and the University of Michigan’s Department of Electrical Engineering and Computer Science. Funding for this pioneering work was generously provided by the National Science Foundation (NSF 2221576), the University of Pennsylvania Office of the President, the Air Force Office of Scientific Research (AFOSR FA9550-21-1-0313), the Army Research Office (ARO YIP W911NF-17-S-0002), the Packard Foundation, the Sloan Foundation, and the NSF National Nanotechnology Coordinated Infrastructure Program (NNCI-2025608), which supports the Singh Center for Nanotechnology, alongside contributions from Fujitsu Semiconductors.

Key co-authors contributing to this research include Maya M. Lassiter, Kyle Skelil, Lucas C. Hanson, Scott Shrager, William H. Reinhardt, and Tarunyaa Sivakumar from the University of Pennsylvania, and Mark Yim from the University of Pennsylvania, as well as Dennis Sylvester, Li Xu, and Jungho Lee from the University of Michigan.