Soterium Earth · RGB
Roadkill Prevention
Fast, affordable warning systems that teach animals to avoid the road, without waiting years for an overpass.
Wildlife overpasses work. They are also expensive, slow to build and limited to a handful of places. Meanwhile, animals cross thousands of ordinary roads every night: county highways, ranch roads, the street in front of someone's house.
Soterium's roadkill prevention work starts from a simple question: what can we put on a car, a driveway or a utility pole this year that warns an animal in a way it actually understands?
Our answer is a family of smart, low-cost deterrents that stay quiet until they are needed, speak to each species in its own sensory language, and change with the seasons so animals never learn to ignore them.
Early-stage concept. The designs below are our own and still to be built and field-tested. We describe the evidence honestly, including what has failed before.
- The idea
- Smart, local deterrents as a fast alternative to overpasses
- Two devices
- A vehicle-mounted system and roadside resonant sentinels
- The principle
- Triggered, species-specific, seasonally rotated
- Why RGB
- Ecology, sensing, electronics and acoustics in one system
The scale of the problem
Every night, on every road
Estimated collisions between vehicles and large animals each year in the United States, counting unreported crashes.
Human injuries each year from reported crashes alone.
People killed each year in reported wildlife-vehicle crashes.
The cost of a single land bridge, up to around AUD 2 million in one Australian estimate, versus tens of thousands for a culvert.
Those figures count mostly deer, elk and other large animals. The toll on smaller wildlife, from foxes and turtles to amphibians and owls, is far larger and mostly uncounted.
Learning from failure
What hasn't worked
Cheap deterrents are not a new idea. Most of them have failed, and the reasons point straight at a better design.
Wrong sound, wrong place
Air-powered deer whistles are considered likely ineffective. Their sound may fall outside the animal's best hearing or be drowned out by road noise.
Too weak to notice
Reviews found little or no effect on collisions. Reflected headlight is faint next to road lighting and the road itself.
Modest at best
A careful trial of light-and-sound units in Tasmania found no statistically significant drop in roadkill, far short of the 50–90% claimed for the device.
Animals learn to ignore them
Any cue that repeats without real danger behind it loses its effect. This habituation is the main reason deterrents fade.
Our approach
Four design rules
Only when it matters
Devices stay silent and dark until an animal and a vehicle are both present. A warning that only appears with real danger is far harder to tune out.
Speak each species' language
Deer see blue and near-ultraviolet light well and orange and red poorly. Hearing ranges differ by species too. Signals are chosen for the animal in front of us, not for human senses.
Rotate with the seasons
Sounds, light patterns and scents change by season and place, so no single cue lasts long enough to become background.
Natural danger first
Research suggests natural warning sounds, such as alarm calls, barking dogs and human voices, trigger stronger flight in deer and wild boar than artificial tones.
Device one
The vehicle system
A compact unit that mounts on the front of a car or truck and watches the road ahead and the roadside within a wide view radius. When it sees an animal heading for the lane, it decides in a fraction of a second how to warn it.
Wide-angle lidar
Lidar maps shape, distance and movement in the dark. Paired with a small thermal camera, it can spot warm bodies in brush before headlights reach them.
Know the animal
An onboard model classifies the species and predicts its path: grazing at the edge, frozen, or about to cross.
A custom warning
The system picks a response for that species: a brief flash at wavelengths it notices and avoids, or a sound pattern played through the horn or a speaker.
Warn the driver too
A chime or dashboard light tells the driver to slow down. Slowing down is still the single most reliable way to prevent a crash.
Map the hotspots
Every sighting is logged with time and place. Across many cars, that becomes a live map of where and when animals cross, data that can guide where roadside devices and crossings go.
- Mounts on
- Cars, trucks, fleet and rural vehicles
- Sees with
- Wide-angle lidar plus thermal imaging
- Responds with
- Species-tuned light, sound and a driver alert
- Bonus
- Crowd-sourced crossing maps
- Placed at
- Driveways, utility and cable poles, known crossings
- Triggered by
- A low-power motion sensor
- Sound
- A tuned resonator, pitched to the target species
- Power
- Tiny solar cell; the ring itself is mechanical
Device two
Resonant sentinels
Not every road needs a computer. Resonant sentinels are simple tuned instruments, like tuning forks or chimes, mounted at driveways and on the utility and cable poles that already line nearly every road.
Each one stays silent until something comes close. A small motion sensor releases a striker, and the sentinel rings with a pure, lasting tone chosen to fall in the target animal's sensitive hearing range. Along a stretch of road, a chain of sentinels becomes a warning system tuned to the animals that live there.
Cheap and durable
A tuned metal resonator has no speaker to wear out and can be made locally at low cost.
Tuned by place
Swap the resonator to match local wildlife: deer in one county, elk or moose in another.
Smarter together
A future version rings only when an animal is near and a vehicle is approaching, the pattern research suggests resists habituation best.
Seasonal rotation
A deterrent that changes with the year
Animal behavior on roads shifts through the year. Each region would get its own rotation library of sounds, light patterns and scents, swapped by season so cues stay fresh.
| Season | What animals are doing | Rotation focus |
|---|---|---|
| Spring | Births; mothers with young; animals drawn to fresh roadside growth | Gentle cues that move young animals away without panic |
| Summer | Young animals dispersing into new territory | Strong, varied cues for naive animals with no road experience |
| Fall | Breeding season; deer moving most and least cautious | Most intense and most frequently rotated cues during peak risk |
| Winter | Migration to winter range; animals drawn to road salt | Cues at crossing points and salt-heavy stretches |
Scent cues are the least proven part of the system: animals get used to synthetic smells quickly. We plan to use them only for short, high-risk windows.
Still to solve
Honest questions
Freeze or flee?
Bright light can make deer freeze instead of run. Light cues must be brief and aimed to move animals away from the lane, not pin them in it.
Panic into traffic
A startled animal can bolt the wrong way. The system must choose when to warn, and when warning the driver alone is safer.
Laws on lights and horns
Vehicle lighting colors and horn use are regulated, and differ by state and country. The design has to stay road-legal.
Neighbors and non-targets
Sentinels must not disturb people, pets, birds or bats, or make nearby habitat less livable.
Proof, not promises
This field is full of products with big claims and weak evidence. Ours must earn trust with real before-and-after data.
The pilot
How we will test it
Find a hotspot
Use state roadkill and crash records to pick a short stretch with frequent collisions.
Count first
Record roadkill and near-misses before anything is installed, ideally across a full season.
Install and compare
Fit sentinels and test vehicles on the pilot stretch, and keep a similar stretch untreated for comparison.
Watch behavior
Use trail cameras to see how animals actually respond: fleeing, freezing, or ignoring the cue over time.
Publish everything
Share the results openly, including what didn't work, so towns and wildlife agencies can decide for themselves.
Keep exploring
Related pages
Sources
- Wildlife-Vehicle Collision Reduction Study: Report to Congress (U.S. Federal Highway Administration, 2008)
- Wildlife deterrents (Biodiversity & Infrastructure Handbook, IENE)
- A trial of a solar-powered, opto-acoustical virtual road-fence to mitigate roadkill in Tasmania (Animals, 2019)
- More clues about how deer see (summary of University of Georgia deer vision research)
- Deer hearing and deterrent research (D'Angelo, University of Georgia, 2007)