Exploración dinámica de fronteras en entornos desconocidos basada en la entropía

[EN] Exploration in disaster areas provides valuable, high-fidelity information to rescue personnel in disaster situations, with the potential to reduce the time for search and recovery of victims. This paper presents an exploration strategy based on entropy to evaluate the frontiers of the known pa...

ver descrição completa

Detalhes bibliográficos
Autores: Godoy-Calvo, Jaime, Lin-Yang, Dahui, Vázquez-Martín, Ricardo, García-Cerezo, Alfonso
Formato: artículo
Fecha de publicación:2023
País:España
Recursos:Universitat Politècnica de València (UPV)
Repositorio:RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Idioma:español
OAI Identifier:oai:riunet.upv.es:10251/192803
Acesso em linha:https://riunet.upv.es/handle/10251/192803
Access Level:acceso abierto
Palavra-chave:Mobile robotics
Autonomous mobile robotics
Robot navigation
Entropy
Information theory
Robótica móvil
Robótica móvil autónoma
Navegación de robots
Entropía
Teoría de la información
Descrição
Resumo:[EN] Exploration in disaster areas provides valuable, high-fidelity information to rescue personnel in disaster situations, with the potential to reduce the time for search and recovery of victims. This paper presents an exploration strategy based on entropy to evaluate the frontiers of the known part of the map using an expectation function. The proposed method employs this metric for exploration planning based on the expectation of future information gain, ensuring a strategy that minimises exploration time while maximising the inclusion of new information into the map. This approach avoids the dependency of the information gain method on fixed-size maps and proposes a sensor-independent model that considers the distribution of obstacles in the frontiers' surroundings. In the evaluation, results are presented in different environments with simulations that demonstrate the efficiency in the exploration planning of the unknown areas, until the complete knowledge of the environment to be explored is completed. The proposed method is publicly available at Godoy-Calvo et al. (2022).