Desvio de obstáculos móveis em andador inteligente por meio das técnicas de desvio de ponto crítico e campos potenciais

Abstract

Smart walkers are assistive devices that employ robotics and electronics in order to expand and enhance the functions that can be performed by traditional orthopedic walkers. One of the most relevant functions of smart walkers is navigations assistance, which includes tasks as path planning and obstacle avoidance. These functions can be useful to people that suffer from vision or cognitive disabilities along with mobility disabilities. This master’s thesis proposes and validates a moving obstacle avoidance technique for smart walkers that intends to improve the way these walkers assist the user’s navigation in environments that are simultaneously being used by other people. The proposed technique – critical point avoidance – attempts to mimic the way pedestrians negotiate space in a corridor by trying to avoid moving obstacles that are in a collision course with the walker in advance, while ignoring obstacles that should not result in a collision. The critical point avoidance technique can be combined with the potential fields technique, which has been shown to perform well when dealing with static obstacles and is also used for obstacle avoidance in smart walkers. The combination of both techniques is an attempt at producing a strategy that is able to better cope with moving and static obstacles. The proposed strategy was validated by comparing, in a simulated smart walker, the classic potential fields strategy to a combination of the critical point avoidance and potential fields techniques. As expected, when simulating a moving object in a corridor, the proposed strategy produced an avoidance action that happens in advance, resulting in weaker avoidance forces and a greater minimum distance to the obstacle.