Edge AI Keeps Wildlife Away: Flox Intelligence Deploys at Railways and Airports

Wildlife encroaching on critical infrastructure like railways and airport runways has long been a thorny problem in global transportation. Traditional loud alarms or fences only work in the short term—animals quickly habituate to these fixed stimuli. Swedish company Flox Intelligence is trying to solve this problem with edge AI: letting devices identify animal species locally and dynamically adjust deterrent sounds to prevent animals from adapting.

How the System Works

Flox's solution consists of battery-powered terminal devices that can be mounted on poles or trees. When motion sensors detect activity, they trigger the camera to capture footage. The device performs animal identification locally—without relying on the cloud—and then emits corresponding sound signals based on the identification results to drive animals away.

The key difference is that the sound isn't static. The system continuously varies the audio patterns it emits, making it difficult for animals to establish a stable association between a particular sound and "harmlessness," thereby avoiding habituation. This logic departs from the "scare them off with louder sounds" approach, shifting instead to "maintain deterrence with varying sounds."

Beyond hardware, Flox also provides a software platform for monitoring device operational status and generating reports. This integrated hardware-software design allows operators to understand which locations are frequently triggered and which species are most active.

Edge AI refers to deploying AI models directly on terminal devices rather than uploading data to cloud servers for processing. Its core advantages lie in low latency, offline availability, and privacy protection: devices perform inference locally without waiting for network round-trips, typically responding within milliseconds. In wildlife deterrence scenarios, edge AI enables the camera to immediately identify and trigger sounds upon capturing an animal, avoiding missed optimal deterrence windows due to network latency. Additionally, since video data doesn't need to be continuously transmitted to the cloud, it saves bandwidth costs and reduces privacy and compliance risks associated with video collection in sensitive areas like airports and railways. Of course, edge devices have limited computing power, so models need to be compressed and optimized, which places higher demands on recognition accuracy.

Habituation is a fundamental concept in animal behavior science, referring to the process by which animals gradually reduce their response intensity to repeated neutral stimuli. In wildlife deterrence, if alarm sounds, flashes, or fences are identical every time, animals quickly learn to associate them with "no actual threat" and stop avoiding them. This is precisely the root cause of why traditional fixed-stimulus solutions fail. The key to countering habituation lies in introducing unpredictability—for example, randomly varying sound frequency, rhythm, and timbre, or alternating calls from different predators, so that animals can never establish a stable "safe" expectation. Flox uses AI to dynamically generate and adjust sound patterns, maintaining deterrence at the behavioral level rather than relying solely on volume.

Deployment and Commercial Progress

Currently, approximately 150 Flox devices are deployed at airports, railway corridors, national parks, and critical infrastructure sites across North America, Europe, and Australia. The company also partners with train manufacturers such as Alstom to integrate its software into vehicle systems. This means future deterrence capabilities may no longer be limited to fixed locations but could move with trains.

In terms of business model, Flox uses a combination of one-time hardware fees plus annual software subscriptions, with subscription fees beginning after the first year. The company projects revenue of approximately $1.5 million in 2026. Other players in this space include Robin Radar, Scarecrow, and iCHASE.

Regarding funding, Flox has raised approximately $5 million from investors including Unconventional Ventures and Tilia Impact Ventures. The company was also a finalist in the Startup Battlefield 200 competition at TechCrunch Disrupt San Francisco and has launched a new $10 million funding round, aiming to expand from 150 active devices to thousands.

Alstom is one of the world's leading rail transit manufacturers, headquartered in France, with business covering high-speed trains, metros, signaling systems, and railway infrastructure. Flox's partnership with Alstom to integrate animal identification and deterrence software into train systems means deterrence capabilities can extend from fixed locations to the front of moving trains. This is particularly important for railway corridors: trains travel at high speeds with long braking distances, so if animals entering the tracks can be detected and deterred ahead of time, collision risks can be significantly reduced. Such partnerships also represent how edge AI is penetrating from standalone hardware into embedded software in large transportation equipment, providing a channel for subsequent large-scale deployment.

Implications of the Technical Approach

Flox's case reflects a broader trend: edge AI is penetrating from consumer electronics into infrastructure scenarios. In environments like railways and airports that are sensitive to latency, network coverage, and privacy, local inference is almost a necessity. Devices don't need to send video to the cloud and wait for instructions—identification and response can be completed in milliseconds, while also avoiding the bandwidth and privacy costs of continuous video transmission.

Another noteworthy point is the "counter-habituation" design philosophy. Many wildlife deterrence solutions fail not because they aren't loud enough, but because they're too predictable. Flox uses AI for dynamic adjustment of sound patterns, essentially countering biological behavioral adaptation. This approach may also be transferable to other human-wildlife conflict scenarios.

Of course, the deployment scale of 150 devices is still limited. In the expansion process from pilot to thousands of units, recognition accuracy, device maintenance costs, and response differences across species will all be tested. The company's projected 2026 revenue also indicates that this remains an early-stage market.

Counter-habituation is not Flox's invention but a long-explored direction in wildlife management. Common methods include: randomizing the time intervals of deterrent signals, rotating playback of multiple predator calls, and combining visual stimuli (such as flashes or moving models). Flox's innovation lies in using edge AI to identify animal species in real time and select or generate corresponding sound patterns accordingly, making deterrence more species-targeted. For example, effective deterrent sounds for birds and deer differ greatly, and playing a single sound uniformly often yields limited results. This "identify—match—vary" closed loop elevates counter-habituation from simple random playback to a data-driven dynamic strategy, offering transferable insights for other human-wildlife conflict scenarios.

One-line Summary

Flox uses edge AI to locally identify animals and dynamically adjust deterrent sounds, countering habituation, with approximately 150 units deployed.

Paragraph Summary

Swedish company Flox Intelligence leverages edge AI to solve the problem of wildlife encroaching on critical infrastructure like railways and airports. Its battery-powered terminals locally identify animal species and dynamically change deterrent sound patterns to prevent animals from habituating to fixed stimuli. Approximately 150 devices are currently deployed at airports, railway corridors, and national parks across North America, Europe, and Australia. The company also partners with train manufacturers like Alstom to integrate software into vehicle systems. Its business model combines one-time hardware fees with annual subscriptions, with projected revenue of approximately $1.5 million in 2026. Flox has raised about $5 million and launched a new $10 million funding round, aiming to expand to thousands of devices. This case reflects the trend of edge AI penetrating into infrastructure scenarios and the design philosophy of using unpredictability to counter biological behavioral adaptation.