首页|基于动态避障风险区域的仿生机器鱼路径规划方法

基于动态避障风险区域的仿生机器鱼路径规划方法

扫码查看
原文链接
  • 国家科技期刊平台
  • NETL
  • NSTL
  • 万方数据
针对来流速度固定、存在多个静态漂浮障碍物和动态障碍物的复杂水环境中胸尾鳍协同推进仿生机器鱼的自主避碰问题，提出了一种基于动态避障风险区域(DAR)的路径规划方法。首先，结合所建立的机器鱼水动力学模型，利用扩展卡尔曼滤波法构造沿障碍物运动方向的类椭球形动态避障风险区域，其长轴与障碍物的运动速度成正比，并通过模糊控制方法对卡尔曼滤波过程的噪声方差进行估计，得到该区域的精确边界;其次，结合机器鱼视场内障碍物的位姿与速度，去除同向运动且速度大于机器鱼的无风险障碍物，实时获得动态环境中的避障风险区域集合，进而得到机器鱼可通行时变区域;最后，根据先避近后避多、边界距离最短原则初步确定先转向、后俯仰的空间避碰策略，进而以障碍物作为外界扰动，设计以期望位姿为输入的非线性模型预测控制器，实时优化得到机器鱼的转弯半径、俯仰角和两侧胸鳍相位差等控制参数，驱动机器鱼安全、快速地通过当前障碍物区域。实验结果表明，机器鱼通过多障碍物区域时，与风险区域边界的最小距离为0。15 m，速度最高达0。15 m/s，空间避障速度最高为0。3 m/s，运动机动性较高且运动轨迹比较平滑，验证了所提方法的有效性。
A Biomimetic Robotic Fish Path Planning Method Based on Dynamic Obstacle-avoidance Risk Region
Aiming at the autonomous collision-avoidance problem of the pectoral and caudal fin co-propelled biomimetic robotic fish in a complex aquatic environment with a fixed incoming flow velocity and the presence of multiple static floating obstacles and dynamic obstacles,a path planning method based on dynamic avoidance risk region(DAR)is proposed.Firstly,an ellipsoid-like dynamic avoidance risk region along the direction of obstacle movement is constructed according to the established hydrodynamic model of the robotic fish by using extended Kalman filtering,whose long axis is proportional to the speed of obstacle movement,and the noise variance of the Kalman filtering process is estimated through a fuzzy control method,to get the precise boundary of the region.Secondly,the non-risky obstacles moving in the same direction as the robotic fish at a greater speed are removed according to the position and speed of obstacles in the field-of-view of the robotic fish,the set of obstacle avoidance risk regions in the dynamic environment is obtained in real time,and then the time-varying passable region of robotic fish is acquired.Finally,a spatial collision-avoidance strategy is preliminarily determined,in which the robotic fish steers firstly and then pitches according to the principles of avoiding the nearest obstacle firstly and then plenty of other obstacles,and keeping the shortest distance from the safe region boundaries.Taking the obstacles as an external perturbation,a nonlinear model predictive controller is designed with the desired pose as input to optimize the control parameters such as turning radius,pitch angle and phase difference between two pectoral fins in real time,so as to drive the robotic fish to pass through the current obstacle area safely and quickly.The experimental results show that when the robotic fish passes through the multi-obstacle area,the minimum distance from the boundary of the risk area is 0.15 m,the speed is up to 0.15 m/s,the spatial obstacle avoidance speed is up to 0.3 m/s,and the spatial maneuverability is high and the trajectory is relatively smooth,which verifies the effectiveness of the proposed method.

biomimetic robotic fishdynamic analysisdynamic obstacle avoidancerisk regionnonlinear model predic-tive control

李宗刚、王治平、夏广庆、康会峰

展开 >

兰州交通大学机电工程学院,甘肃兰州 730070

兰州交通大学机器人研究所,甘肃兰州 730070

大连理工大学工业装备结构分析优化与CAE软件全国重点实验室,辽宁大连 116024

北华航天工业学院河北省跨介质飞行器重点实验室,河北廊坊 065000

展开 >

仿生机器鱼 动力学分析 动态避障 风险区域 非线性模型预测控制

国家自然科学基金国家自然科学基金国家自然科学基金国家自然科学基金国家自然科学基金国防基础科研计划甘肃省高等学校产业支撑计划大连理工大学工业装备结构分析国家重点实验室开放课题兰州交通大学军民融合创新团队培育基金

6166302012175032121020821227504412211530449JCKY2022407C0092022CYZC-33GZ22119JMTD202211

2024

10.13973/j.cnki.robot.230281

机器人
中国自动化学会　中国科学院沈阳自动化研究所

机器人

CSTPCD北大核心
影响因子：1.134
ISSN：1002-0446
年,卷(期)：2024.46(4)