RAPID/Collaborative Research: Data Collection for Robot-Oriented Disaster Site Modeling at Champlain Towers South Collapse

This Grant for Rapid Response Research (RAPID) will support a collaborative team of researchers from Florida State University, Texas A&M University, and Carnegie Mellon University to operate under the supervision of the Miami Dade Fire Rescue Department and Florida Task Force 1 at the site of the Champlain Towers South condominium collapse in Surfside, Florida. This project will address the need for a robot-oriented model of rubble by using unmanned aerial system (UAS) imagery and other contextual information.

The lack of a robot-oriented model of rubble is a major barrier to the design and manufacture of effective, economical, and reliable ground robots for disasters and other extreme environments. Although structural engineering teams are also investigating the site, they do not capture data about the factors that impact whether a robot can navigate the interior of a building collapse.

This project will benefit society by facilitating the design and deployment of robots to save lives, either to find survivors in rubble otherwise inaccessible to humans and dogs or by reducing the need for human responders to enter unsafe areas. The team is diverse, with a woman as the principal investigator, and will train a diverse set of students to conduct robotics research for disasters.

The team will: 1) assist rescue, recovery, and forensic structural teams by collecting UAS images of the collapse from response through recovery, 2) collect and analyze data on UAS performance relating to flights, missions, data processing, and operations tempo, 3) analyze orthomosaic and digital elevation imagery to formally model traversability constraints for ground robots in extreme environments, including features such as scale, shape, and surface properties, 4) curate images for general use and archive on the Texas Data Repository open source dataverse site, and 5) attempt to create a 3D visualization of the voids in the interior of the rubble from the progressively uncovered site via a subtractive and labeling process. The research will create a new fundamental research methodology for analyzing disasters, and extreme environments in general, from the perspective of ground and aerial robotic systems.

The image datasets may also enable the computer vision machine learning communities to recognize structural conditions and indications of survivors. The results of the study will be made freely available, including a workshop, and will improve use of robots in future disasters by formalizing design features and offering a rapid recognition of which robot types to deploy for what conditions.

This project is supported by the cross-directorate Foundational Research in Robotics program, jointly managed and funded by the Directorates for Engineering (ENG) and Computer and Information Science and Engineering (CISE).

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.