The situation

Firefighters regularly face extreme risks: chemical threats, collapsing structures, and zero-visibility conditions. In such environments, robotic assistance can save lives by extending human capabilities. Meaningful human-robot collaboration must be first grounded in the way firefighters actually work: their team structure, communication, and decision-making under stress.

My role, team and responsibilities

I worked directly with two senior research engineers and interacted with several city stakeholders and firefighting employees for driving the research forward.

As a research engineer, I supported the generative research and led the experimental prototyping phase. My goals were to:

1. Help better understand firefighters' communication, collaboration and deployment model, and

2. Explore how robotic agents could support them in real-world firefighting workflows and hostile environments. 

Observing, listening and analyzing

We visited fire stations and conducted in-situation interviews to understand how firefighters make decisions and communicate during crisis. We also observed a live simulation of a soccer stadium fight, involving firefighters, safety guards, and paramedics. Hundreds of volunteers were acting as soccer fans who came to watch the game with some having fake injuries for the sake of the exercise.

We first learned and observed the deployment of a pyramidal command structure, where teams of two firefighters operate under a chief, and chiefs themselves form hierarchies as team size increases. This insight directly informed system scalability and information flow.

We then quickly discovered significant discrepancies in how disciplines reported casualties and conditions, pointing to the need for better ways and tools to align teams around a shared understanding of the crisis:

• Lack of consistent terminology or mental models across emergency teams

• Limited technological integration in high-pressure, low-visibility contexts

Building a research-informed proof of concept

Our observational research insights combined with existing research on human-machine cooperation provided me with opportunities to build a proof of concept of a firefighter-robots cooperation system. This POC was very preliminary and attempted to demonstrate the feasibility and scalability of such concept.

The system included a modular robot control interface with intuitive toggles and feedback streams to be used by firefighters, Mindstorm robots equipped with infrared, ultrasound and video sensors, and a WIFI/bluetooth communication protocol to pair them.

I provided firefighters with the ability to increase and decrease the number of robots needed based on the scenario at hand. They had access to video and sensor feedback from the robots (e.g., infrared, ultrasound), levels of robot automation (manual, semi-autonomous, autonomous modes) and joystick control, on-demand communication channels and tasks to be performed.

The impact

• Identified critical gaps in shared situation awareness during emergency responses provided meaningful directions for firefighters communication training roadmap.
  
  

• The proof of concept sparked strong interest among firefighters in the practical applications of robotic support tools and was handed off to a graduate researcher for formal user testing and further development and supported the foundation of a Ph.D. thesis.