← Back ICRA 2026

Towards Distributed Robotic Casualty Assessment Using Multimodal, Non-Contact Perception and Probabilistic Inference

Zachary Bortoff, Srijal Shekhar Poojari, Kleio Baxevani, Joshua Gaus, Christopher Titus, Ahmed Ashry, Derek Paley

PDF

AI summary

Key figure (auto-extracted from paper)

Fusing multimodal non-contact sensor data via probabilistic inference enables reliable, distributed robotic triage in noisy, real-world disaster scenarios.

Robotic triage Multimodal fusion Probabilistic inference Mass-casualty incidents Non-contact sensing Dynamic Bayesian networks

Problem

Mass-casualty incidents overwhelm medical personnel, yet existing robotic triage systems struggle with noisy environments, lack of sensor fusion, and inability to operate asynchronously across distributed platforms.

Approach

The system equips ground and aerial robots with cameras, microphones, and radar to collect non-contact data, then combines these heterogeneous streams using a dynamic Bayesian network to estimate casualty injury states.

Key results

Modality-specific classifiers for injury detection and vital sign estimation
Dynamic Bayesian network enabling distributed, asynchronous sensor fusion
Reliable casualty state estimation in timed mock field trials
Significant accuracy gains from modeling injury correlations over independent baselines

Why it matters

Provides emergency responders with scalable, automated triage support during large-scale disasters where human assessment is overwhelmed or unsafe.

Abstract

Mass-casualty incidents demand rapid and accu- rate triage, but the scale and acuity of injuries often overwhelm available medical personnel. To address this, we present a system that enables ground and aerial robots to localize and assess casualties using non-contact sensors, including color and thermal cameras, millimeter wave radar, and microphones. Injury and vital sign measurements from modality-specific classifiers are fused using a probabilistic model that captures correlations between injury states and supports distributed, asynchronous evidence accumulation. We validate the system through a series of timed mass-casualty field experiments using custom-built drones and Boston Dynamics Spot ground robots customized for robotic medical triage, demonstrating reliable estimation of casualty states and robustness to noisy conditions and sensor drop out.

Index terms

Field Robots Search and Rescue Robots