Research on state-parameter estimation of unmanned Tractor—A hybrid method of DEKF and ARBFNN
Guangfei Xu,Meizhou Chen,Xiangkun He,Yulong Liu,Jian Wu,Peisong Diao
DOI: https://doi.org/10.1016/j.engappai.2023.107402
IF: 8
2023-11-04
Engineering Applications of Artificial Intelligence
Abstract:Unmanned tractor relies on multi-sensor information collection to obtain the current state or parameters. However, when driving in complex field, it will inevitably suffer from the uneven ground, the impact of crop straw and clods et al., which usually poses considerable challenges for multi-sensor to obtain stable and accurate value of states and parameters to realize tractor automatic lane guidance control. Therefore, a novel state and parameter estimation method by mixing the dual extended Kalman filter (DEKF) technology and adaptive radial basis function neural network (ARBFNN) technology is proposed in this paper. Firstly, DEKF technique is applied to estimate key states and initial model time-varying parameters–front/rear axle cornering stiffnesses at the same time. Then, in order to further improve the accuracy of estimation value of front/rear axle cornering stiffnesses during the automatic lane guidance control process, an ARBFNN technology is investigated by taking the heading error, lateral error and initial estimation value of front/rear axle cornering stiffnesses from DEKF as inputs to approach ideal estimation value. Finally, results of automatic lane guidance control scenarios from both simulation and hardware-in-loop (HIL) implementation show that the proposed hybrid estimated method of DEKF and ARBFNN can robustly obtain satisfactory estimation value and automatic lane guidance control performance for tractor when the control system is characterized by both time-varying model parameters and uncertain external energy-bounded disturbance. A comparative study is also conducted to investigate cases to show its effectiveness when the hybrid DEKF-ARBFNN state and parameter estimation method is used and when it is not.
automation & control systems,computer science, artificial intelligence,engineering, electrical & electronic, multidisciplinary