Real-time Obstacle Avoidance of Hovercraft Based on Follow the Gap with Dynamic Window Approach
Yuanhui Wang, Wenchao She, Mingyu Fu, Fuguang Ding, Shaoshi Dai · 2018
Considering the high-speed hovercraft navigation safety and its complicated maneuvering and control performance, it is necessary to study the real-time obstacle avoidance combined with the hovercraft dynamics performance. This paper introduced an improved Follow the Gap Method (FGM) with dynamic window approach (DWA) considering the real-time hovercraft dynamic performance to avoid obstacles. The traditional DWA is used to find an appropriate velocity, that is the vehicle selects the possible velocity in the admissible velocity space to avoid collision during a specific time interval. And then choose the most appropriate velocity by using an objective function. In addition, this method mostly applies in vehicles with first-order nonholonomic constraints such as automobile and has the problem of performance decline applying in hovercraft. For the improved DWA applied on hovercraft, the acceleration and deceleration limits are added to the algorithm and the influence of the choice of heading angle during collision avoidance process is more emphasized. In this paper, follow the gap method is used to find reference heading angle. FGM aims to find the optimal command angle by selecting maximum gap in hovercraft's view and combining the maximum gap center azimuth with goal azimuth using a heuristic and fusing function. Traditional FGM has the problem of calculating difficulties in nonholonomic constraints and ignoring the working scope of radar. The improved FGM addresses these problems by restricting rudder angle and classifying the obstacles. In this way, the collision avoidance goal is achieved by properly guiding the heading angle and leading an accessible velocity. Effectiveness of this combined algorithm as well as the parameter regulation strategy is verified on a dynamic simulative model of hovercraft.