APPLICATION OF NEURAL NETWORKS FOR PATH PLANNING OF AN AUTONOMOUS ROBOT

Petrov Denis Sergeevich

the application of modern mobile robotics and various types of autopiloted transportation requires efficient motion planning algorithms. Such an algorithm should build both a global trajectory and take into account local terrain parameters and state parameters of the controlled object, sensor indicators. Neural networks and machine learning are one of the promising areas that can realize an algorithm with the above requirements. The paper analyzes the algorithms for planning the movement of autonomous mobile objects on the basis of neural networks, the features of each method are considered. Classification of these methods by the level of control and the amount of data used about the environment and the state of the object. Based on the analysis, an approach is selected for further research of solving the problem of efficient motion planning.

Keywords:path planning, mobile robotics, neural networks, machine learning, reinforcement learning, deep learning