TY - GEN

T1 - Improved neural network 3D space obstacle avoidance algorithm for mobile robot

AU - Tong, Yuchuang

AU - Liu, Jinguo

AU - Liu, Yuwang

AU - Ju, Zhaojie

PY - 2019/8/3

Y1 - 2019/8/3

N2 - Path planning problems are classical optimization problems in many fields, such as computers, mathematics, transportation, robots, etc., which can be described as an optimization problem in mathematics. In this paper, the mathematical model of obstacle environment is established. The characteristics of neural network algorithm, simulated annealing algorithm and adaptive variable stepsize via linear reinforcement are studied respectively. A new neural network 3D space obstacle avoidance algorithm for mobile robot is proposed, which solves the problem of the computational duration and minimum distance of the traditional neural network obstacle avoidance algorithm in solving the optimal path. According to the characteristics of the improved neural network algorithm, it is fused with a variety of algorithms to obtain the optimal path algorithm that achieves the shortest path distance and meets the requirements of obstacle avoidance security. The simulation experiment of the algorithm is simulated by Matlab. The results show that the improved neural network spatial obstacle avoidance algorithm based on the multiple algorithms proposed in this paper can effectively accelerate the convergence speed of path planning, realize the minimum path distance, and achieve very good path planning effect.

AB - Path planning problems are classical optimization problems in many fields, such as computers, mathematics, transportation, robots, etc., which can be described as an optimization problem in mathematics. In this paper, the mathematical model of obstacle environment is established. The characteristics of neural network algorithm, simulated annealing algorithm and adaptive variable stepsize via linear reinforcement are studied respectively. A new neural network 3D space obstacle avoidance algorithm for mobile robot is proposed, which solves the problem of the computational duration and minimum distance of the traditional neural network obstacle avoidance algorithm in solving the optimal path. According to the characteristics of the improved neural network algorithm, it is fused with a variety of algorithms to obtain the optimal path algorithm that achieves the shortest path distance and meets the requirements of obstacle avoidance security. The simulation experiment of the algorithm is simulated by Matlab. The results show that the improved neural network spatial obstacle avoidance algorithm based on the multiple algorithms proposed in this paper can effectively accelerate the convergence speed of path planning, realize the minimum path distance, and achieve very good path planning effect.

KW - global path planning

KW - obstacle avoidance algrithm

KW - improved neural network algorithm

KW - adaptive variable stepsixe

KW - simulated annealing

U2 - 10.1007/978-3-030-27538-9_10

DO - 10.1007/978-3-030-27538-9_10

M3 - Conference contribution

SN - 978-3-030-27537-2

T3 - Lecture Notes in Computer Science

SP - 105

EP - 117

BT - Intelligent Robotics and Applications

A2 - Yu, Haibin

A2 - Liu, Jinguo

A2 - Liu, Lianqing

A2 - Ju, Zhaojie

A2 - Liu, Yuwang

A2 - Zhou, Dalin

PB - Springer

ER -