MilliMobile Robot Moves Without a Battery — Here’s How

Estimated reading time: 10 minutes

Key Takeaways

• The MilliMobile robot is smaller than a penny, weighs less than a paperclip, and it never needs a battery.
• MilliMobile is the first gram-scale microrobot that can move, sense its environment, run its own computing, and send data wirelessly — all powered entirely by harvested light and radio frequency (RF) energy.
• MilliMobile has been tested on 10 different surfaces, in low indoor lighting, and on RF signals so weak they are barely detectable.
• The key insight behind MilliMobile is intermittent motion, using capacitors to store small bursts of harvested energy and release it in short pulses.
• In the world of smart agriculture, infrastructure inspection, defense, and industrial IoT, this changes what is possible.

The MilliMobile robot is smaller than a penny, weighs less than a paperclip, and it never needs a battery. Not a rechargeable one. Not a replaceable one. None at all.

Built by researchers at the University of Washington, MilliMobile is the first gram-scale microrobot that can move, sense its environment, run its own computing, and send data wirelessly — all powered entirely by harvested light and radio frequency (RF) energy from the air around it. It was presented at the ACM MobiCom 2023 conference in Madrid, one of the most respected venues in mobile computing research.

This is not a lab trick that only works under perfect conditions. MilliMobile has been tested on 10 different surfaces, in low indoor lighting, and on RF signals so weak they are barely detectable. It still moves. It still communicates. It still senses.

In the world of smart agriculture, infrastructure inspection, defense, and industrial IoT, this changes what is possible. Let’s look at exactly how it works, what makes it special, and where it is headed.

What Is the MilliMobile Robot and Why Does It Matter for Battery-Free Microrobots?

Most small robots face the same problem: batteries. They are heavy, they die, and in large deployments, replacing them costs enormous amounts of time and money. For swarms of hundreds or thousands of tiny sensing robots spread across a farm, a factory floor, or a remote pipeline — battery maintenance alone can make the entire idea impractical.

MilliMobile solves this from the ground up.

The robot’s chassis measures exactly 10 x 10 millimetres. Its total mass is under 1.1 grams. That is smaller than most insects. Despite that size, it carries a solar cell on top, a Bluetooth chip, temperature and humidity sensors, four light-direction sensors, and two small vibration motors — all working together without a single battery cell.

The key insight behind MilliMobile is intermittent motion. Instead of moving continuously like a traditional motor-driven robot, it uses capacitors to store small bursts of harvested energy — between 47 and 150 µF — and then releases that energy in short pulses to drive its vibration motors. The wheels roll a little. The capacitors charge again. Repeat.

This approach means the robot can operate using less than 100 µW of power — a level so low that indoor light or ambient RF signals can sustain it indefinitely.

Here is what makes that remarkable in context: 100 µW is roughly 10,000 times less power than a typical smartphone consumes while the screen is on. MilliMobile does useful sensing and communication at that level, continuously, with no battery ever needed.

Sources:

• University of Washington news release: University of Washington news release
• ACM MobiCom ’23 technical paper: ACM MobiCom ’23 technical paper

The Engineering Behind Light and RF Powered Robots at This Scale

Understanding MilliMobile means understanding its three core engineering choices: its body, its power system, and its communication hardware.

The Body

The chassis is laser-machined from carbon fibre and polyimide laminate — two materials chosen because they are stiff, light, and thin. FR4 material (the same stuff used in standard circuit boards) forms the three-spoke wheels, which provide 3 mm of ground clearance. That clearance matters. It lets the robot roll over small pebbles, gaps in concrete, and uneven terrain without getting stuck.

The whole structure is flat and compact, meaning the solar cells on top can capture light from as many angles as possible.

The Power System

MilliMobile harvests energy from two sources:

1. Light — Thin-film solar cells made by Microlink Devices sit on top of the robot. These cells can generate enough power even under under-cabinet indoor lighting. In outdoor conditions or brighter indoor environments, they generate significantly more. The robot can begin moving at illuminance levels as low as 20 W/m².
2. RF energy — A 908 MHz rectifier circuit captures radio frequency energy from the air. At a ceiling-height transmitter setup of about 2.4 metres, this delivers tens of µW — enough to start motion when light is unavailable. The robot can cold-start (begin working from zero power) at just 7 µA, which is an impressively low threshold that makes it viable in dim or signal-weak environments.

Both sources feed into the capacitor bank. The robot’s on-board controller decides when enough charge has built up to trigger a burst of motion or a sensor reading.

The Communication System

MilliMobile uses the nRF52 Bluetooth SoC (system-on-chip), which is a well-known ultra-low-power wireless chip. Here is what it achieves in practice:

• BLE (Bluetooth Low Energy) advertising packets sent every 5 seconds at an average current draw of only ~11 µA
• Long-range coded PHY packets received at up to 700 metres from a base station (with approximately 9% packet delivery rate, which is usable for periodic environmental sensing)
• Robot-to-robot peer communication at up to 140 metres using coded PHY

A 0.5 dBi antenna on the robot communicates with an 8 dBi antenna on the base station. The asymmetry is intentional — you put the more powerful antenna at the fixed base, and keep the robot’s antenna tiny and lightweight.

The robot also has four photodiodes arranged around its body to detect the direction light is coming from. This is how it navigates — it turns toward the brightest light source, using light itself as a passive navigation signal. No GPS. No camera. No complex mapping algorithm. Just phototropism, like a plant, but in robot form.

Performance Numbers

• Maximum speed: ~5.5 mm/s (about 19.8 metres per hour in ideal conditions)
• Tested speed over 60 minutes: 10.8 metres covered
• Payload capacity: Up to 3 times the robot’s own weight — roughly 3 grams — with only a 25% speed reduction at a 1-gram additional load
• Surfaces tested: 10 types including compact soil, linoleum, concrete, carpet, and wood

Source:

• ACM MobiCom ’23 technical paper: ACM MobiCom ’23 technical paper

🌱 Your Business Still Runs on Manual Work. It Doesn’t Have To.

Stop spending hours on tasks that software can handle in seconds.

Whether you run a small business or manage complex operations, manual workflows quietly eat your most valuable resource — time. Business automation removes repetitive work from your plate so your team can focus on what actually grows the business.

👉 See How Flowgrammer Automates Your Business (Flowgrammer Automates Your Business)

Autonomous Sensing Swarms — The Real-World Applications

A single MilliMobile robot moving 10 metres per hour is interesting. A thousand of them, deployed across a field or factory, communicating with each other and with base stations — that is where the vision becomes genuinely powerful.

Precision Agriculture

Modern farms need constant data: soil moisture, temperature, humidity, and microclimate variation across large areas. Today, this usually means fixed sensor stakes — expensive to deploy, limited in coverage, and dead once their batteries die.

A swarm of MilliMobile-style robots could crawl slowly across a field, powered by sunlight, continuously reporting temperature and humidity data back to a base station. They would never need battery replacement. They could navigate toward the brightest patches of a crop canopy or spread across a dark greenhouse using RF harvesting from a WiFi router.

The University of Washington team specifically identified precision agriculture as one of the primary target markets for this technology. The robot’s HDC2010 sensor already reads temperature and humidity — exactly the data farmers need.

Industrial and Infrastructure Inspection

Factories, railways, roads, and bridges all need regular inspection. Much of this inspection today is done manually — a human walks a route, checks equipment, logs readings. That is slow, expensive, and sometimes dangerous.

MilliMobile’s ability to operate in RF-rich electromagnetic environments is particularly relevant here. Industrial floors are full of radio signals from manufacturing equipment, WiFi routers, and machine sensors. Rather than interfering with those signals, MilliMobile can harvest them. The more connected the factory, the more energy the robot has to move.

AFCEA’s 2024 defense and technology analysis described MilliMobile as a “next big thing in swarm robotics,” specifically calling out its suitability for infrastructure inspection and hazardous industrial environments where human presence is risky and battery servicing is impractical.

Defense and Battlefield Sensing

This is the application that defense analysts find most compelling. In contested or hazardous areas, deploying a human to check sensors or replace batteries is a serious tactical and safety problem. A swarm of battery-free microrobots that can be released, navigate using available RF energy or light, and send back environmental or signals-intelligence data over hundreds of metres — that concept directly addresses a real operational need.

Because MilliMobile operates at such low power levels and is physically tiny, it also presents a much smaller detectability signature than traditional sensor platforms.

Source:

• AFCEA signal media analysis: AFCEA signal media analysis
• ACM MobiCom ’23 technical paper: ACM MobiCom ’23 technical paper

What the Research Community Says About This Platform

The ACM MobiCom ’23 paper — published under DOI 10.1145/3570361.3613304 — positions MilliMobile as solving a problem that has blocked microrobotics for years. Previous battery-free robots could either move or sense or communicate. MilliMobile is the first to demonstrate all three simultaneously at gram scale in realistic conditions.

The research team deliberately tested the robot outside of carefully controlled lab setups. Moving across carpet, rough concrete, and compact soil are not clean laboratory tests. They are the kinds of surfaces a real deployment would encounter.

The funding behind the project also signals its broader ambitions. The work was supported by National Science Foundation Graduate Research Fellowships and a NASA Space Grant — both programs tied to long-term U.S. federal priorities around technology development, sustainability, and reducing reliance on conventional power sources.

The environmental angle is also explicitly named in the paper itself. Lithium batteries carry significant environmental costs in production and disposal. At massive IoT scale — billions of sensor nodes — that adds up to an enormous material footprint. Battery-free microrobots like MilliMobile represent one credible path toward distributed sensing infrastructure that does not depend on the lithium supply chain.

Source:

• University of Washington news release: University of Washington news release
• ACM MobiCom ’23 technical paper: ACM MobiCom ’23 technical paper

The Limits and What Comes Next

It is important to be clear-eyed about what MilliMobile is and is not right now.

Current limitations include:

• Speed — 5.5 mm/s maximum is very slow. For covering large areas quickly, you would need many robots working in parallel, not just one.
• Navigation complexity — The robot navigates toward light. That works well in some environments, but complex indoor spaces, tunnels, or underground infrastructure present harder navigation challenges.
• Packet delivery at range — The 9% delivery rate at 700 metres is usable for infrequent environmental sensing, but it would not support time-sensitive or high-data applications.
• Payload and terrain — While the robot handles 10 terrain types well, very rough outdoor terrain, wet surfaces, or steep inclines remain untested at this scale.

What the research community expects next:

• More efficient energy harvesting circuits, pushing the operating threshold even lower
• More sophisticated on-board navigation, possibly using multiple photodiode readings to build simple local maps
• Swarm coordination protocols where robots share data with each other to cover more ground
• Application-specific sensor payloads — gas detectors, vibration sensors, or acoustic sensors for specific industrial use cases

MilliMobile is a platform and a proof of concept. The most important thing it demonstrates is that the fundamental combination — battery-free power, sensing, computing, and wireless communication at gram scale — is achievable in realistic conditions. The next generation of work builds on that foundation.

Sources:

• ACM MobiCom ’23 technical paper: ACM MobiCom ’23 technical paper
• AFCEA signal media analysis: AFCEA signal media analysis

Why This Matters for the Future of IoT and Smart Infrastructure

The Internet of Things has a battery problem. Analysts estimate there are already over 15 billion connected IoT devices in the world, and projections for the coming decade push that number dramatically higher. Every one of those devices needs power. Most use batteries. Replacing, recycling, and managing those batteries at scale is already a logistical and environmental challenge — and it will get harder as deployments grow.

MilliMobile represents a different model: sense everything, power nothing deliberately.

Instead of shipping batteries to sensors, you design sensors that live on whatever energy the environment already provides — light, radio signals, even thermal gradients. The sensor works as long as the environment does.

For smart farms, that means sensors that outlast the farm itself. For factories, it means maintenance-free sensing nodes embedded in floors and walls. For infrastructure monitoring, it means putting sensors in locations where no one can easily reach to change a battery — inside concrete pillars, under bridges, along remote pipelines.

The research was clear in 2023. The ACM MobiCom community recognized MilliMobile as a genuine step-change in what battery-free robotic platforms can do. The practical deployments are still coming, but the underlying capability is proven.

Source:

• University of Washington news release: University of Washington news release

Frequently Asked Questions