This dataset was obtained after experiments in which a small drone needed to autonomously explore an indoor environment in search of a target. The mathematical framework used in this research allowed us to model decision-making under uncertainty. The mission was modelled following a classic Search and Rescue operation, where the agent needed to explore a cluttered environment in low visibility and with no GPS. The target used was a heated mannequin, and its position was unknown to the drone. The sensor used for target detection was a thermal camera.  

The dataset consists of simulation results, where smoke was used as a visual obscurant, and real-life testing results, where low light was used as a low visibility condition. It contains CSV files with the output of the decision-making algorithm (observations, actions...), and rosbags with the drone position, occupancy map of the environment, beliefs states, and odometry.