Intelligent UAV Search in Uncertain Environments
AI-powered guidance when the search space is large and time is critical
Overview
Lightweight AI for real-world autonomy
Aeris-UK supported our US sister company, Aeris LLC, in developing a lightweight, intelligent search system for UAVs, tailored to their government client’s needs.
The solution enables platforms to rapidly locate hard-to-find, uncertain targets – using AI-guided route planning that runs entirely on-device.
The Challenge
​How do you search when you don’t know exactly what – or where – you’re looking?
Conventional search patterns waste time and energy when information is scarce and environments are vast. The client needed a system that could guide a UAV towards the most likely areas of interest or concern – quickly, adaptively and without access to cloud infrastructure.
Existing methods couldn’t deliver, especially on mobile devices operating in denied or disconnected settings. The core challenge: deliver real-time, intelligent search under tight computational constraints.
Our Approach
Smarter search, smaller footprint.
We developed RADA – a modular framework for exploring and benchmarking AI-guided search strategies. It supported a range of approaches including rule-based, probabilistic and machine-learned techniques, all tested under consistent, real-world conditions.
To avoid dependence on proprietary data, we built a surrogate simulation engine to produce representative data and used it to train lightweight machine learning models that could replicate complex planning outputs with minimal compute. These trained models were deployed as a microservice, optimised for mobile platforms and field integration.
