Quadrupedal Robots in Search and Rescue: Perception and Teleoperation

• Autores: Christyan Mario Cruz Ulloa
• Directores de la Tesis: Antonio Barrientos Cruz (dir. tes.)
• Lectura: En la Universidad Politécnica de Madrid ( España ) en 2024
• Idioma: inglés
• Tribunal Calificador de la Tesis: Arturo de la Escalera Hueso (presid.), Francisco Javier Badesa Clemente (secret.), Mario Andrei Garzón Oviedo (voc.), Carlos Balaguer Bernaldo de Quirós (voc.), Alfonso José García Cerezo (voc.)
• Programa de doctorado: Programa de Doctorado en Automática y Robótica por la Universidad Politécnica de Madrid
• Calificación: Sobresaliente Cum Laude
• Materias:
  • Matemáticas
    • Ciencia de los ordenadores
      • Inteligencia artificial
    • Investigación operativa
      • Sistemas de control
  • Ciencias tecnológicas
    • Tecnología de los ordenadores
• Enlaces
  • Tesis en acceso abierto en: Archivo Digital UPM
• Resumen

  • Search and rescue robotics emerges from the need to assist brigades in tasks involving significant danger, mainly during early explorations. Post-disaster events such as earthquakes or attacks tend to destroy buildings and cause material and human losses. First aid brigades are those in charge of carrying out early interventions to identify and rescue unconscious or trapped victims. However, this work implies exposing his life to potential structural collapses or electrical, toxic risks. Artificial intelligence systems, embedded systems, specific sensors and robots with particular modes of locomotion have been part of the modern development experienced in the last decade. These tools have become valuable instruments for developing expert systems and carrying out complex missions focused on the search and rescue field. Although there are indeed a series of challenges within search and rescue robotics, this thesis seeks to address four fundamental axes, having as protagonists quadruped robots equipped with intelligence systems, specific detection sensors, and robotic manipulators to detect and assist victims in these conditions. The primary axis concerns locomotion and progression across unstructured terrains, commencing with terrain interpretation and characterization and culminating in generating walking patterns and sophisticated obstacle-surmounting strategies. The second axis is dedicated to victim identification through various ranges of the visible spectrum invisible to the human eye, such as thermal and multispectral systems combined with neural networks. Detection has been integrated with a third axis, which addresses an intelligent search system based on the behaviour of a first-responder, aimed at establishing a robust model for active search and victim identification. Finally, the fourth axis explores teleoperation through immersive interfaces for quadruped robots and legged-manipulator-type robots, alongside the management of environmental information. Tests have been carried out in reconstructed environments, both indoors and outdoors, to validate each part that makes up the thesis, which includes drills with rescue teams from the UME (Military Emergency Unit); All of this has made it possible to generate a series of results and contributions to the state-of-the-art, which have been reflected in publications in international journals and conferences, which support this doctoral thesis.