Dynamic obstacles avoidance algorithms for unmanned ground vehicles

Abstract

In the recent years, unmanned ground vehicles (UGVs) are being increasingly used as service robots. Unlike industrial robots, which are situated in fixed and controlled positions, UGVs work in dynamic environments, sharing the space with other vehicles and humans. UGVs should being able to move without colliding with any obstacle, assuring its integrity and the environment safety. In the state of the art, navigation algorithms for UGVs are able to plan routes in a safe way, with static obstacles and work in partially controlled environments. However, when the environment is dynamic, the paths planned are more dangerous and often result in more energy and resources consumption, or even can block the UGV in a local minima situation. In this thesis, adaptation of state of the art algorithms for working in dynamic environments has been proposed. These algorithms take into account time information, such as based on curvature arcs (PCVM and DCVM) and based on dynamic window approach (DW4DO and DW4DOT). A global path planning algorithm based in Lattice State Planner (DLP) that can solve situations where an obstacle avoidance algorithm does not work is also proposed. These algorithms have been validated in both simulated and real tests using several robotic platforms, including an assistant robot (RoboShop) that has also been designed and built during this thesis development.