Learning-Based Platooning Control of Automated Vehicles Under Constrained Bit Rate

Platooning control is a promising scheme to alleviate traffic congestion while maintaining the safety of road traffic. Vehicles' cooperation deeply depends on the capability of communication infrastructure suffering from bandwidth constraints. This paper focuses on the platooning control problem of a class of nonlinear automated vehicles under constrained bit rate. A neural network (NN)-based observer is constructed to estimate the vehicular state, where a coding-decoding scheme is presented to alleviate the communication burden and an NN with carefully designed tuning laws is adopted to approximate the unknown nonlinearity. By using the Lyapunov stability theory, a sufficient condition is received for realizing the required formation with the given constant inter-vehicle spacing. Based on this condition, the desired gains of NN-based observers and decoder-based controllers are accessible by resolving matrix inequalities insensitive to the number of vehicles. Furthermore, the upper bound of platoon tracking errors dependent on bit rate is analytically derived. Finally, an illustrative example shows the validity of the developed control scheme.