3D Path Planning In Dynamic Environment Based On Improved Fruit Fly Optimization Algorithm

With the rapid development of modern high technology,robot technology innovation has gradually become an important research trend in the academic field.Robots can replace human beings to complete work in the special environment,to effectively reduce labor intensity and reduce the error rate.Path planning is the key research content of mobile robot controlling and the research goal is to find a shorter,safer and smoother path from the starting point to the ending point according to the robot's moving environment.At present,the traditional optimization algorithm and swarm intelligence optimization algorithm are two main kinds of algorithms to solve the path planning problem.Different algorithms have their own advantages,but there are also various deficiencies.In this thesis,to solve the robot path planning problem in the static environment and dynamic environment,the artificial potential field method is added to the fruit fly optimization algorithm,which makes the two algorithms complementary to each other.The main research contents are as follows:(1)The step size determines the search efficiency and convergence speed of the algorithm.In the traditional fruit fly optimization algorithm,the step size is a fixed value,so the environment information is not fully utilized resulting in many unnecessary searches of the algorithm in the open local area and many illegal individuals in the dense area.In this thesis,a kind of adaptive search step determined by environmental information is proposed.According to known obstacle information,drosophila population adjusts the step size in real time at each search.The adaptive step size can not only overcome the shortcomings of the fixed step to the algorithm,but also endow the algorithm with higher obstacle avoidance capability.In the simulation experiment,the numbers of iteration of the two algorithms with different step sizes circumventing the different sizes of obstacles are compared.From the results,the number of iteration of the algorithm with adaptive step size is reduced by nearly 50%.(2)In the fruit fly algorithm,the smell concentration determining function is used to evaluate the individual quality of the population and the smell concentration determining function directly affects the safety and length of the final path.In the traditional fruit fly algorithm,the smell concentration determining function only depends on the distance from the individual to the ending point and does not take into account the safety of the individual,which leads to contact between the path and obstacles.Because the robot is abstracted as a particle in the simulation experiment,the path ultimately searched is a legal path,but the robot in the real environment has a certain size and shape,the path will be not conducive to the safe movement of robots.Based on the above-mentioned situation,in this thesis,the distance between individuals and obstacles is used to evaluate the safety of individuals,and the farther distance means the individual is safer.Finally,the convex combination of safety factors and distance factors is used as the smell concentration determining function,and the coefficients of the convex combination are determined by the local environment of drosophila population.(3)The correction process is that the algorithm adjusts the path searched to improve the overall quality of the path.In this thesis,an improved artificial potential field method is proposed as an amendment to amend the path when the distance between the path and the obstacle is less than the presupposed safety value.In order to overcome the locking phenomenon of the artificial potential field,in addition to calculating gravitational force and repulsion force,a pulling force generated by other individuals is added and the coefficient of pulling force depends on the smell concentration of the individual.The higher the concentration is,the greater correction of the path to the individual.Finally,the joint position of the three forces is used to correct the population position in real time.Through simulation experiments,the influence of main parameters on the final path is analyzed,then the best choice of parameters is determined.Finally,the algorithm is tested in the static and dynamic environment,and the experimental results are compared with the ant colony algorithm,particle swarm optimization and evolutionary programming algorithm,which shows the effectiveness and superiority of the proposed algorithm.