A Semantic World Model for Urban Search and Rescue Based on Heterogeneous Sensors

In urban search and rescue scenarios, typical applications of robots include autonomous exploration of possibly dangerous sites, and the recognition of victims and other objects of interest. In complex scenarios, relying on only one type of sensor is often misleading, while using complementary sensors frequently helps improving the performance. To that end, we propose a probabilistic world model that leverages information from heterogeneous sensors and integrates semantic attributes. This method of reasoning about complementary information is shown to be advantageous, yielding increased reliability compared to considering all sensors separately. We report results from several experiments with a wheeled USAR robot in a complex indoor scenario. The robot is able to learn an accurate map, and to detect real persons and signs of hazardous materials based on inertial sensing, odometry, a laser range finder, visual detection, and thermal imaging. The results show that combining heterogeneous sensor information increases the detection performance, and that semantic attributes can be successfully integrated into the world model.