智能轮式机器人在养殖场中路径规划的研究

发布时间：2018-05-05 06:35

本文选题：移动机器人 + 路径规划　；参考：《长春大学》2017年硕士论文

【摘要】：在机器人研究领域中,路径规划问题一直都是研究的重点。近年来,随着机器人技术和产业的飞速发展,对路径规划的研究受到越来越多的学者、专家的关注和重视。现今,对路径规划方法的研究已取得丰硕的成果,但是也存在着不足。许多算法的提出只是在实验仿真的基础上进行的,并不能用于实际情况中。这些不足需要研究者们去进一步完善。所以本文结合吉林省教育厅项目“大型养殖场监控机器人控制系统设计”,对养殖场中机器人路径规划进行研究。首先,针对静态环境下的机器人路径进行研究,提出了一种改进的遗传算法。在该方法中,一是改进了地图环境的建立,将预先设定的静态障碍物直接引入到算法的初始种群中,避开了环境建模问题;二是在算法中设置了检查装置,保证了生成的新个体不在障碍物内;三是改进了适应度函数的设计,考虑到机器人路径的最短距离、路径平滑度、安全性能(避开障碍物),将这三个因素加入到适应度函数的设计中;四是对适应度函数中的三个因素人为加入权值系数,进一步确保得出最优路径。通过仿真实验,表明该方法是可以实现的。其次,对基于改进遗传算法的机器人路径规划方法进行优化研究,针对遗传算法中适应度函数的权值,提出利用粒子群优化算法进行自主寻优,避开了人为设置的不足。根据遗传算法适应度函数中三个权值的关系,设计出粒子群优化算法的适应度函数。该方法能够自行确定各个因子的权值,实现权值因子的自主协调,从而得到最优路径。经过仿真实验,表明该方法可行。再次,运用蚁群算法对机器人进行路径规划仿真研究。仔细分析了蚁群算法的原理、参数和基本公式模型。通过仿真实验得出该算法的可行性。最后,针对养殖场的实际环境,对机器人路径规划进行模拟仿真研究。以养鸡场环境为模型,根据机器人在养鸡场中执行任务的不同,分别研究了监控模式(巡航模式)和路径规划模式。机器人在不同的工作模式下,路径规划方法不同。监控模式下采用了步长法,路径规划模式下采用了遗传算法和蚁群算法。根据得到的遗传算法和蚁群算法路径规划仿真结果进行对比,得出遗传算法更适用于养鸡场环境下的机器人路径规划,相较于蚁群算法,遗传算法更具有优越性。
[Abstract]:In the field of robot research, path planning has always been the focus of research. In recent years, with the rapid development of robot technology and industry, more and more scholars and experts pay attention to the research of path planning. Nowadays, the research on the path planning method has made a lot of achievements, but there are still some shortcomings. Many algorithms are proposed only on the basis of experimental simulation, and can not be used in the actual situation. These deficiencies need to be further improved by researchers. So this paper studies the path planning of the robot in the breeding farm with the project of Jilin Provincial Education Department "Design of the Control system of the Monitoring Robot in the Large-scale breeding Farm". Firstly, an improved genetic algorithm is proposed to study the robot path in static environment. In this method, one is to improve the establishment of the map environment, the other is to introduce the pre-set static obstacles directly into the initial population of the algorithm, to avoid the environmental modeling problem, and to set up a checking device in the algorithm. Thirdly, the design of fitness function is improved, considering the shortest distance of robot path, path smoothness, Safety performance (avoiding obstacles and adding these three factors to the design of fitness function) and adding weights to three factors in fitness function to ensure that the optimal path can be obtained. The simulation results show that the method is feasible. Secondly, the robot path planning method based on improved genetic algorithm is optimized. Aiming at the weight of fitness function in genetic algorithm, the particle swarm optimization algorithm is proposed for autonomous optimization, which avoids the deficiency of artificial setting. According to the relation of three weights in the fitness function of genetic algorithm, the fitness function of particle swarm optimization algorithm is designed. This method can determine the weight value of each factor and realize the independent coordination of the weight factor, thus the optimal path can be obtained. The simulation results show that this method is feasible. Thirdly, the ant colony algorithm is used to simulate the path planning of the robot. The principle, parameters and basic formula model of ant colony algorithm are analyzed in detail. Simulation results show the feasibility of the algorithm. Finally, the robot path planning is simulated and simulated according to the actual environment of the farm. Based on the model of chicken farm environment, the monitoring mode (cruise mode) and the path planning mode were studied according to the different tasks performed by the robot in the chicken farm. Under different working modes, the path planning method of robot is different. Step size method is used in monitoring mode, genetic algorithm and ant colony algorithm are used in path planning mode. According to the simulation results of genetic algorithm and ant colony algorithm, it is concluded that genetic algorithm is more suitable for robot path planning in chicken farm, and genetic algorithm is more superior than ant colony algorithm.
【学位授予单位】：长春大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】