A compact, AI-powered system for rapid decontamination during CBRNE incidents

Abstract

Chemical, Biological, Radiological, Nuclear and Explosives (CBRNE) incidents have long posed significant challenges to humanity, continuously testing our ability to respond timely and effectively to the complex threats they present. The response phase of the disaster  anagement cycle is crucial for mitigating immediate harm and preventing further contamination, especially in such high-risk scenarios. (Coppola, 2011) During this phase, decontamination is vital to neutralize or remove hazardous agents from affected individuals and prevent cross-contamination and the spread of contaminants to unaffected areas. (Lake et al., 2013) However, despite its importance, decontamination methods often face considerable challenges.

Traditional decontamination systems demand extensive infrastructure, significant setup times and specialized personnel, making them impractical for rapid deployment in urgent situations. Moreover, decontamination techniques, including physical removal, disinfection, sterilization and ultrasonic cleaning, often struggle with scalability and adaptability to different types of contamination. (Dennis, et al., 2023) The lack of real-time body posture monitoring of the affected individuals and the needed manual guidance leads to inconsistencies in coverage, undermining the effectiveness of emergency response efforts. (Carter and Amlôt, 2016) Meanwhile possible cultural and linguistic barriers between the end-users and the first  responders can cause further confusion and lack of communication in such high-stress scenarios. (Carter et al., 2012) Despite its importance, current decontamination mechanisms often struggle with operational constraints, including resource limitations and the effectiveness of handling diverse CBRNE threats. Overcoming these limitations is essential to ensure faster, more efficient and safer decontamination during the response phase.

Addressing these  challenges, this Work-in-Progress (WiP) paper presents: i) the Fast Deployable Mass Decontamination system (FDMD), a novel decontamination tunnel designed for rapid deployment, operational flexibility and ease of use in various emergency response environments, and ii) the Decontamination Body Pose Estimation Tool (DBPET), a camera-based image-analysis system that tracks body posture in real-time, while handling the workflow and guidance of the end-users during the process, integrated with the FDMD to ensure optimal decontamination coverage.

The FDMD is a modular, lightweight decontamination tunnel that can be rapidly deployed and assembled, supporting both wet and dry decontamination techniques through three sets of interchangeable sprinklers. Its compact design enables efficient storage within a standard suitcase, facilitating easy transport, deployment and prepositioning. To optimize decontamination body coverage and reduce human error, the FDMD is integrated with the AI-driven DBPET. This self-sustaining small-scale battery-powered module utilizes real-time camera-based image analysis to track body posture and continuously guide individuals to ensure that they maintain the correct pose for effective decontamination. The tool provides automated, multilingual, step-by-step audio instructions, enabling clear guidance even in high-stress situations. By leveraging artificial intelligence, the system enhances compliance with international decontamination standards and ensures thorough exposure to decontamination agents.

Preliminary evaluation of the system focuses on key performance indicators, including deployment speed, body posture efficiency and operational scalability. Initial results suggest that the FDMD can be assembled in under ten minutes, significantly reducing response time compared to traditional systems. DBPET has demonstrated the potential to enhance user compliance with decontamination protocols, leading to more effective contaminant removal. By offering a cost-effective, rapidly deployable and adaptable solution, the system has the potential to improve emergency response efforts, providing a viable alternative to existing large-scale decontamination infrastructures.

This paper presents an overview of the system's architecture and workflow, components, sub-systems, and key functionalities, while examining its potential impact and benefits in emergency response operations. The system’s impact is assessed across several dimensions: economic, societal, operational, technological and environmental. Example scenarios will be also examined and presented to showcase the system’s versatility and effectiveness in various emergency contexts and environments. These include large-scale CBRNE incidents, industrial accidents and localized environmental emergencies, highlighting the system’s capacity to improve response time, reduce contamination risks and enhance safety for both responders and affected individuals.

Future work will involve further performance testing under controlled conditions and real-world deployment scenarios to assess the system’s effectiveness. Other testing procedures will include studies on the maintenance of the different components of both FDMD and DBPET. Additional research will focus on fully automating the decontamination process through DBPET handling the mechanical valves of the FDMD and monitoring of the sprinklers’ consumables (e.g. decontaminants levels), increasing the AI-based assistance by deploying further useful models (e.g. sentiment detection), integrating FDMD and DBPET with other response technologies such as triage tagging systems for additional security and real-time updates and evaluating its applicability in industrial and civilian settings.

By addressing current limitations in decontamination procedures, the FDMD along with DBPET represent a significant advancement in emergency response, enhancing the safety and efficiency of decontamination efforts for both experts and non-experts managing hazardous incidents.