Predictive Follow the Gap Method for Dynamic Obstacle Avoidance

摘要

Advancements in autonomous mobile robots hinge on refining key components like mapping and path planning to address identified limitations. The local planner, crucial for obstacle avoidance, is a component of path planning. The Follow the Gap Method (FGM) stands out as a simple and effective obstacle avoidance algorithm. FGM calculates possible passage points by assessing gap sizes and positions of obstacles. Our focus lies in enhancing FGM's adaptability to dynamic environments. Introducing Predictive FGM, we incorporate robot and dynamic obstacle data to forecast future gaps and obstacle states. By integrating predictive elements, the algorithm selects gaps based on anticipated changes, enabling safer navigation by predicting the states of gaps and obstacles when they are closest to the robot. Evaluation via Monte Carlo simulations and real-world experiments with an autonomous wheelchair in dynamic environments show the effectiveness of Predictive FGM over standard FGM.