Real Time Underwater Obstacle Avoidance and Path Re-planning Using Simulated Multi-beam Forward Looking Sonar Images for Autonomous Surface Vehicle

Abstract

Abstract. This paper describes underwater obstacle avoidance and path re-planning techniques for autonomous surface vehicle (ASV) based on simulated multi-beam forward looking sonar images. The sonar image is first simulated and then a circular obstacle is defined and created in the field of view of the sonar. In this study, the robust real-time path re-planning algorithm based on an A* algorithm is developed. Our real-time path re-planning algorithm has been tested to regenerate the optimal path for several updated frames with a proper update frequency between the start point and the goal point both in static and dynamical environments. The performance of proposed method is verified through simulations, and tank experiments using an actual ASV. While the simulation results are successful, the vehicle model can avoid both single obstacle, multiple obstacles and moving obstacle with the optimal trajectory. For tank experiments, the proposed method for underwater obstacle avoidance system is implemented with the ASV test platform. The vehicle is controlled in real-time and moderately succeeds in its avoidance against the obstacle simulated in the field of view of the sonar together with the proposed position stochastic estimation of the vehicle.

On the 3D Track Planning for Electric Power Inspection Based on the Improved Ant Colony Optimization and A∗ Algorithm

“…e ant tends to preferentially choose a track with a higher pheromone concentration until it can find the optimal track from its nest to the food source. e A * algorithm [25,26] combines the characteristics of Dijkstra and BFS algorithms. It considers both the points close to the starting point and the points close to the target point so that it can find the shortest route length between any two points when there are obstacles.…”

ACO-A*: Ant Colony Optimization Plus A* for 3-D Traveling in Environments With Dense Obstacles

“…Existing path planning algorithms can be roughly classified into two types, the deterministic ones and the nondeterministic ones. There are popular deterministic algorithms, such as fast marching [2], mixed integer linear programming [10], A* search algorithm [11], and A*-based dynamic algorithms, e.g., sparse A* search [12]. These algorithms are often effective to find optimal solutions on the premise of a good environment modeling.…”

“…Thus, we design a length-based ratio rl to reduce the estimated cost in such case so as to better estimate the optimal path cost. Mathematically, the edge cost of e ij is set as (11) where cost(r i , r j ) is the optimal traveling cost from r i to r j which can be obtained from the representative cost map, cost(i, r i ), and cost(r j , j) have been obtained during representative selection and they are generally equal to the traveling costs of the optimal path(i→r i ) and path(r j →j), respectively. Thus, the sum of these three cost values approximates the optimal traveling cost of path(i→r i →r j →j).…”

“…The New Zealand Region bathymetric datasets (2016) are free available online: https://www.niwa.co.nz/our-science/oceans/bathymetry [2017-[10][11][12][13][14][15][16][17][18][19]]. …”