Search Theory in the Age of AI
By Doug Lothian, Chief Science Officer, International Association of Search and Rescue Coordinators (IASARC)
If you fall into the water, even with modern aircraft, sensors, and highly trained rescuers searching for you, the odds of survival remain dauntingly low.
Fewer than 19% of searches for persons in the water end in a life saved, and when flotation or signaling devices are absent, the probability drops dramatically. The central reason is sobering: humans, despite their dedication and equipment, are not reliable detectors in the dynamic maritime environment. A U.S. Coast Guard study (2006) showed that even when an aircraft flew within 0.1 nautical mile—just two football fields—of a target, the probability of detection was only 18%.
Search Theory: A Proven Science
The discipline of Search Theory was developed in World War II to bring mathematical rigor to the challenge of submarine hunting and has since provided the foundation for search planning in SAR. Its principles remain central to modern operations, yet while the science is robust, the practical application is still limited by human variability.
At its simplest, two conditions define a successful search:
The object must be detectable by the sensor.
The sensor must recognise and identify the object as the target.
Simple in theory, yet a complex series of actions and interactions must take place to meet these conditions.
First, there needs to be awareness that an emergency situation is or may be occurring. Today we have modern communication systems that can alert search professionals to your predicament and location, from anywhere on the globe.
Once aware, the SAR facility gathers additional information and begins planning the search. This involves understanding the location and type of object in distress (e.g. person, vessel, aircraft etc), the weather on scene, the sea conditions, tides, currents, available search resources etc. Through complex simulation, drift models are created to predict movement over time. Operational Search Plans are formed based upon the available search resources in the area – only then can the SAR facilities be sent to the area to commence the Operational Stage. It is during the operational stage, assuming all previous actions have been successful, that the object may enter the field of view of the Search Observer Sensor for detection.
While detection can be thought of as a mathematical problem, generally dictated by geometry – height and range / distance of the target from the observer - recognition and identification is more nuanced and complex.
The Human Challenge
Many scientific studies have sought to solve the problem of recognition or “predicting the probability of target discrimination” including the Johnston Criteria, NVESD Targeting Task Performance (TTP) metric, BAE ORACLE, FLIR92 (Rand/Bailey) Classical Model of Search and Sarnoff Visual Discrimination Model. All assume a human observer will perform the final recognition and identification stage, once a sensor presents a possible target.
Detection and recognition is never guaranteed. Numerous survivor accounts describe aircraft, helicopter or vessels passing directly overhead yet failing to detect and recognise them.
Humans may spot something unusual - they may differentiate between a life raft and a marine mammal, a person from a crab pot marker, or a buoy. However, every human is different – different experiences, training, physical attributes and state of mind. Indeed, the International Aeronautical and Maritime Search and Rescue Manual (IAMSAR) describes the situation succinctly:
“the eye is vulnerable to the vagaries of the mind. We can “see” and identify only what our mind permits us to see.”
Physical limitations are well documented; the eye struggles with rapid refocusing, requiring one to two seconds to switch between near and distant objects. In environments where no fixed reference points exist – like the open ocean - , the eye may fail focus altogether, a condition known as empty field myopia. Even within the eyes central field of view, resolution falls sharply just 20o off center.
Meeting the two conditions of a successful search is therefore a phenomenally complex task.
How can we make improvements?
Today, new technologies are reshaping this landscape.
Sophisticated planning tools, modeled on the U.S. Coast Guard’s Search and Rescue Optimal Planning System (SAROPS), are being adopted worldwide and allow more efficient deployment of finite SAR resources.
Advanced sensors such as Electro-Optical Infrared (EO/IR) and Synthetic Aperture Radar (SAR) are expanding detection into poor visibility and challenging conditions.
Artificial intelligence and machine learning are demonstrating the ability to detect and track persons and objects in the water with consistency and precision beyond human capability.
These innovations offer the potential for a step-change in SAR effectiveness, moving beyond the constraints of human-in-the-loop detection.
Yet a fundamental challenge remains: the lack of a common framework to measure, compare, and validate these technologies. Without internationally agreed standards, the SAR community cannot confidently assess which tools deliver genuine improvements in survival outcomes.
IASARC’s Role: Global Standards
The International Association of Search and Rescue Coordinators (IASARC) is responding by working with global experts in search theory, operational practices, sensor development, and AI to establish an internationally accepted set of requirements and verification and validation (V&V) metrics.
Through its new Science & Technology Workstream, IASARC is creating a rigorous, evidence-based framework for evaluating both legacy and next-generation technologies. Key objectives include:
A global forum for open exchange of scientific and technological lessons learned.
A peer-reviewed SAR Science Bibliography and Data Hub for research and operational data.
Technology-neutral performance standards and repeatable test protocols for planning tools, AI detection software, and modern sensors.
Benchmark datasets and intercomparison trials to enable fair, reproducible evaluation.
Support for open-source planning and detection tools so that even resource-constrained organisations can benefit from cutting-edge science.
The goal is simple but powerful: ensure that investments in planning tools, sensors, and AI systems deliver measurable improvements in detection, survivability, and mission efficiency worldwide.
Looking Ahead
Search Theory gave SAR its scientific foundation. AI and modern sensors now offer the chance to assist and potentially overcome the limits of human observation.
But technology must be matched by trusted standards and shared data if it is to save more lives.
IASARC’s mission is to advance the profession of SAR Coordination through advocacy for improved law and regulation, scientific advancement, and rigorous professional certifications. Ultimately this helps ensure every SAR practitioner—wherever they operate—has access to validated, effective tools and the knowledge to use them.
Notable Links
To support IASARC’s lifesaving mission now | Donate
Follow IASARC | LinkedIn
