A Fusion Method of Local Path Planning for Mobile Robots Based on LSTM Neural Network and Reinforcement Learning

Na Guo,Tengteng Gao,Di Wang,Guoming Liu,Yongdi Li,Caihong Li,Mohammad Amin Hariri-Ardebili

doi:10.1155/2021/5524232

Na Guo, Tengteng Gao + Show 5 more

Open Access

https://doi.org/10.1155/2021/5524232

Copy DOI

Abstract

Due to the limitation of mobile robots’ understanding of the environment in local path planning tasks, the problems of local deadlock and path redundancy during planning exist in unknown and complex environments. In this paper, a novel algorithm based on the combination of a long short-term memory (LSTM) neural network, fuzzy logic control, and reinforcement learning is proposed, and uses the advantages of each algorithm to overcome the other’s shortcomings. First, a neural network model including LSTM units is designed for local path planning. Second, a low-dimensional input fuzzy logic control (FL) algorithm is used to collect training data, and a network model (LSTM_FT) is pretrained by transferring the learned method to learn the basic ability. Then, reinforcement learning is combined to learn new rules from the environments autonomously to better suit different scenarios. Finally, the fusion algorithm LSTM_FTR is simulated in static and dynamic environments, and compared to FL and LSTM_FT algorithms, respectively. Numerical simulations show that, compared to FL, LSTM_FTR can significantly improve decision-making efficiency, improve the success rate of path planning, and optimize the path length. Compared to the LSTM_FT, LSTM_FTR can improve the success rate and learn new rules.

Highlights

Robots are widely used in industry, agriculture, medicine, military, and other fields, and can assist or replace human work
In previous work [42], we proposed a path planning algorithm (FL) based on fuzzy control, which can be used for data collection. is paper uses both unsupervised and supervised learning with collected data to pretrain the model and optimizes the model through reinforcement learning
Based on the average decision-making time in Figure 20(b), the real-time decision-making performances of LSTM_FT and LSTM_FTR with the same neural network structure are similar in time performance; they are markedly better than the fuzzy logic control method. e neural network model can accelerate the calculation speed and improve the operating efficiency of the system

Summary

Introduction

Robots are widely used in industry, agriculture, medicine, military, and other fields, and can assist or replace human work. Global path planning calculates an optimal route based on the given start and target points while avoiding obstacles in a known environment [5, 6]. Global path planning primarily includes A∗ search algorithm, rapidly exploring random tree (RRT) and Voronoi diagram algorithm [7, 8]. They typically require long computation times, when the map is larger. Local path planning explores a collision-free optimal path to reach the target point based on environmental information detected by onboard sensors [9, 10]. To ensure that the robot can avoid obstacles safely and reach the target point faster, it is necessary to investigate algorithms in more detail [14, 15]

Methods

Results

Conclusion