Application of Deep Neural Network in AGV Real Time Navigation Optimization
In order to improve the autonomy and intelligence of AGV navigation control,a hybrid intelligent real-time optimal control method based on deep neural network(DNNS)is proposed.The obstacle avoidance problem of AGV trajectory planning is transformed into a nonlinear optimal control problem(OCP).The smooth transformation is used to deal with the path con-straints,and the Gaussian pseudo spectral(GPM)method is used to obtain the optimal solution of the problem.On this basis,a real-time optimal controller based on DNNS is designed to realize the optimal trajectory planning of AGV.The results of numeri-cal experiments show that the optimal control method based on DNNS can generate the optimal control instructions to guide AGV to the target position,and has high robustness to the initial conditions.It can meet the constraints of different obstacles,and real-ize the real-time navigation optimization strategy of AGV.
Deep LearningNeural NetworkAGVNonlinear Optimal ControlTrajectory Planning
NETL
NSTL
万方数据
