Comparative evaluation of path planning algorithms in a simulated disaster environment

Anticipating and managing the human flow is essential for saving human lives during a disaster. Unmanned Aerial Vehicle (UAV) helps in human detection and navigating to the location of the injured person within a short time. This paper compares Q-learning and state-action-reward-state-action (SARSA) path planning algorithms for guiding the UAV to the desired location. A virtual disaster environment is created using Gazebo and the path planning algorithms are executed in the UAV using the Robot Operating System (ROS) platform. The UAV is equipped with a SONAR sensor for obstacle avoidance. For the experiment, the UAV starts and ends at fixed locations. The experiment results show that the Q-learning algorithm performs better than the SARSA algorithm.