Calibrated, provenance-aware decision support for urban search and rescue.
Exploring how to show responders what is known, what remains uncertain, and where each piece of information came from.
Mission
To shorten the path from uncertain information to better-supported action.
The problem
After a building collapse, responders must decide where to commit limited teams, time and equipment while working with incomplete, conflicting and changing information.
Information can arrive from different sources, at different times and with different levels of reliability. When provenance, recency and uncertainty are separated from the information itself, prioritisation becomes harder to reason about and harder to explain afterwards.
What ARESTOR is exploring
A software evidence layer for worksite assessment and prioritisation. It would combine information responders already have, such as reports, observations and imagery, into one picture that:
States its uncertainty: expresses how strongly the available evidence supports relevant operational hypotheses, including possible live-victim presence.
Shows its provenance: where each piece of information came from and how recent it is.
Updates as new information arrives, so the picture changes as the evidence changes.
SourcesReports, observations, imagery
→
Evidence pictureUncertainty, provenance, recency
→
DecisionMade by the responder
Responders keep the decision. ARESTOR is intended to support operational judgement, not replace it. The current work is software-first and does not autonomously task response teams or equipment.
Current status
Early-stage research and validation. ARESTOR is not a deployed operational product and has not been validated for operational use.
Speaking with search-and-rescue practitioners about how worksite prioritisation decisions are made under uncertainty.
Planning structured decision exercises to test whether probabilistic information changes prioritisation behaviour.
Technical work on calibrated belief updating and provenance-aware evidence representation, to date on synthetic scenarios, with testing on real disaster imagery as the next step.
Who I would like to hear from
Practitioners in urban search and rescue, technical search, command and coordination, and emergency management, for short conversations.
Teams and agencies interested in structured tabletop exercises.
Researchers working on uncertainty, information fusion and decision support in disaster response.
Founder
ARESTOR is being developed by Leon Long, a former Singapore Armed Forces explosive ordnance disposal officer with a BS in Computer Engineering and Mechanical Engineering from Boston University. LinkedIn