WPI Researcher Introduces Batlike Rescue Drones

At disaster sites where cameras crack and GPS fails, a team at Worcester Polytechnic Institute is putting its money on biology. Led by roboticist Nitin J. Sanket, the research team has developed palm-sized drones that fly based on echolocation, imitating how bats “see” with sound to dart through smoke and dust in situations like search and rescue.

Why bats are the blueprint for resilient micro rescue drones

Rescuers frequently need to see where people cannot. Visual sensors get blinded by particles and low light, while lidar can be bulky and power-sucking for tiny flyers. Chirping as you go, some bats solve a similar problem biologically: They send out chirps, listen to the echoes, and assemble their aural map so quickly that they can catch bugs at warp speed. It’s a strategy, Sanket argues, that suits the constraints of micro-robots weighing less than one gram in terms of power and onboard compute.

The WPI prototypes rely on ultrasound and AI in the onboard computer to categorize echo reflections, estimating objects within about a two-meter radius. Rather than use costly custom transducers, the team began with the same low-power ultrasonic emitters used in automatic faucets—parts engineered to sip power and work dependably for years.

Seeing through smoke and debris when cameras fail

In a collapsed building and wildfire smoke, cameras have a short life span, and there is also clutter from hot spots even with thermal imagers. Echolocation filters away a great deal of that noise. Agencies such as FEMA’s Urban Search and Rescue task forces and the test facilities at the National Institute of Standards and Technology have long emphasized the need for small, insistent scouts that could crawl into voids and report back with no tethered power or pristine visibility.

Public safety use of small drones has exploded in recent years, according to the Center for the Study of the Drone at Bard College, yet most deployed platforms still rely on vision-first navigation. An acoustic-first microdrone bridges the gap, as it works where vision-first autopilots do not—indoors and underground.

Taming propeller noise with bat-inspired acoustic shaping

Early prototypes have faced a challenge: Many small rotors produce broadband noise, which can mask the robot’s own chirps and scramble the sensor.

The solution was straight out of the anatomy of the bat. Many leaf-nosed species mold their calls—they’re sculptable—thanks to a variety of flexible tissues around their noses and ears, tuning and aiming sound in the air on the fly. Buoyed by that success, the WPI team created a 3D-printed acoustic shaper deployed at the front of the drone to sift through and steer ultrasound out of rotor clutter into the environment.

Software does the rest. Lightweight neural networks for separating self-noise from environmental echoes operate onboard in real time, removing the need for cumbersome processors. (Vievere says this also drains compute and the battery as little as possible, which is key in drones that have to be both small and cheap.) The result is a cleaner acoustic image.

From moonshots to near-term impact in search and rescue

Sanket’s journey to bats began with another bio-inspired vision: swarms of micro-drones cross-pollinating crops. The idea was a moonshot away from the innumerable constraints of outdoor flight and scale. Also, search and rescue offered a more immediate target at which biology’s toolkit — efficient sensing, minimal compute, robust navigation — could potentially make a difference soon, with metrics for testing it much clearer and agency partners that had more defined needs.

How a small rescue robot thinks in tight, noisy spaces

The vehicles emit ultrasonic pings, then infer range and bearing to the nearest obstacles, building a coarse 3D occupancy map within a couple of meters. At this scale, short planning horizons are sufficient. Think of a bat scanning instant to instant rather than plotting out an entire map. The autonomy stack focuses on collision avoidance and target-finding, delegating higher-level mapping and victim detection to a connected ground station or bigger drone where possible.

What success in the field could look like for responders

Imagine a warehouse fire that is billowing dark smoke as part of it collapses. A responder throws a small cluster of micro-bots through an open door; they spread out, weave between the shelves, and report back a crude map and cautionary flags — blocked aisles, unexplained drops in terrain, moving heat sources — for human teams outside. The two-meter bubble was calculated to be sufficient to keep them from falling apart while still being useful, while the low mass reduces risk if they touch any victims or equipment.

Next hurdles for the bat bots: speed, endurance, and tinyML

Speed and endurance are still the big problems. Faster flight requires faster sensing and planning without bloating the processor or battery. The group is investigating better acoustic beamforming, improved rotor isolation, and even event-driven neural networks that rouse themselves only when echoes change — all in line with the trend toward “tinyML,” which aims to keep inference in the milliwatt range.

Regulatory and validation pathways are clearer than in the past. NIST keeps standardized obstacle courses for small robots, and several fire departments now operate live-burn training sites where uncrewed systems can be tested against authentic smoke and heat. They’re the kinds of environments that will help shape the hardware and autonomy before any deployment with first responders.

Part of a bioinspired wave shaping practical robotics

Batlike drones are only part of a wider push to learn lessons from nature, including the insect-scale RoboBee at Harvard and Bat Bot at Caltech that emulated flapping-wing flight dynamics in a soft-bodied robot. What sets the WPI work apart is its pragmatism: a rotorcraft platform, off-the-shelf parts, and bat-inspired acoustics to find their way when vision falls painfully short. It’s biology as a systems manual, not drapery.

If it works as designed, the payoff is simple: more survivable micro-robots that go to places humans can’t, buy precious moments when seconds count, and bring a little of a bat’s superpower closer to the people who need it most.