Finding Moving Operators in Firefighting Operations Based on Multi-Goal Next-Best-View Exploration

Abstract

Many recently designed exploration algorithms for search and rescue are based on the expansion of existing exploration algorithms to multiple simultaneously operating agents. These algorithms are quite useful in extensive search and rescue operations but neglect the need for adaptability necessary in small scale environments. Therefore, this paper proposes a novel modular multi-layer approach, which combines conventional Next-Best-View Exploration with predefined boundary conditions to enable a multi-goal driven search for victims and operators. The boundary conditions are mapped on cost maps individually and fused dynamically in a common weighting matrix. Exemplary conditions are the last known operator pose or estimated positions of fire sources. The exploration algorithm compares nearby points of interest in regards to their weight and chooses an appropriate navigation goal. The method is evaluated for usage in context of firefighting operations with teams of humans and robots.