The Hybrid Navigation Method in Face of Dynamic Obstacles

摘要

With the development of society and lifestyle changes. Robotic navigation based on traditional navigation and positioning techniques can hardly cope with navigation in complex dynamic scenarios. Traditional navigation and positioning algorithms, such as Dijkstra and A*, cannot perform path planning in dynamic scenarios. In this paper, we present an improved reinforcement learning-based algorithm for local path planning that allows it to still perform well when there are more dynamic obstacles. The algorithm has the advantage of low cost to transform into others situation and fast training speed over the navigation algorithm based entirely on reinforcement learning.