| Tractor-Trailer robot is a typical intelligent field engineering robot.It can realize autonomous navigation in field area unstructured environment by relying on environmental perception,trajectory planning and tracking control without driver participation or a small amount of intervention.It is an effective tool to realize unmanned navigation in the field.Based on the unmanned skidding vehicle in forest area,taking the Tractor-Trailer autonomous navigation robot as the research object,and aiming at realizing the intelligence and unmanned of the Tractor-Trailer robot,its unstructured road recognition and tracking control method is studied in the special field environment.Using the self-developed TractorTrailer intelligent navigation platform and multi-sensor navigation system,the relevant theoretical research is carried out,and the feasibility of the proposed navigation strategy is verified by experiments.The research on environmental perception and tracking control of Tractor-Trailer intelligent navigation platform is completed in the field unstructured road environment.The main research contents,methods and conclusions of this study include:(1)Considering the complex field environment and the special structure of Tractor-Trailer robot,according to the structural characteristics of existing unmanned skidding vehicles,an intelligent TractorTrailer robot platform is designed and developed.The overall design of the platform includes chassis design,hardware layer,system layer and application layer.The chassis design mainly includes the structural design of the Ackerman chassis tractor and four-wheel unpowered Trailer.The hardware layer design includes the selection and installation of on-board computer,gyroscope,2D-dimensional lidar and monocular camera.The system layer mainly refers to the selection and construction of the main control computer.The Linux system(Ubuntu 16.04)and the ROS kinetic(Robot Operating System)are installed on the main control computer.The establishment of the platform will provide the basis for the subsequent verification of road recognition algorithm,the test of system control method and the development of experiments in indoor and outdoor environment.(2)In order to realize unstructured road recognition in complex field environment,based on the multi-sensor system carried on the Tractor-Trailer robot platform,an unstructured road recognition algorithm based on the combination of 2D-dimensional lidar and monocular vision camera is developed.Compared with the traditional urban structured road and rural semi-structured road recognition algorithm,this algorithm can better adapt to the field complex environment and the recognition of unstructured road.The experimental results show that the advantage of this strategy is that it can not only better meet the real-time and accuracy requirements of the Tractor-Trailer robot in the unstructured road environment in the field area,but also the lightweight code can improve the operation efficiency of the vehicle main control computer,reduce the hardware cost of the platform and improve the reliability of the equipment in the harsh environment in the field area.(3)According to the special structure of the Tractor-Trailer robot,this study proposes the kinematic model of the Tractor-Trailer robot at low speed.Based on this model,the cooperative trajectory planning algorithm of the Tractor-Trailer robot is further developed.In this method,the phenomenon that the trailer body deviates to the inner side of the curve in the turning process of the Tractor-Trailer system is no longer simply regarded as the tracking error,but the phenomenon is fully analyzed and evaluated.The inherent characteristics of the system are determined.In this study,it is regarded as the inherent attribute of the system,and the corresponding motion law is analyzed and summarized.The establishment of motion model and the development of motion planning algorithm show that the trajectory planning of trailer system is significant and predictable under general road conditions.Therefore,this method has the advantages of taking into account and balancing the control performance of the tractor and the tracking stability and trajectory smoothness of the trailer.(4)Based on the cooperative trajectory planning of the Tractor-Trailer robot system,a model predictive control method(MPC)is proposed.Through the simulation experiment and indoor experiment of the trailer system,the cooperative trajectory planning algorithm and tracking control strategy proposed in this study are verified.The vehicle test results show that the predicted results are in good agreement with the experimental results.The actual lateral tracking error of the intelligent Tractor-Trailer platform in the indoor environment shall not exceed 0.15 m,and the lateral tracking error after the system is stable in the unstructured road environment in the field area shall not exceed 0.45 m.The feasibility of the tracking control strategy proposed in this study is verified.The experimental results show that the system based on the proposed method can achieve stable tracking,and the control effect is better than the traditional method.The Tractor-Trailer robot system and field autonomous navigation strategy proposed in this study provide a theoretical and technical cornerstone for the intellectualization of Tractor-Trailer autonomous operation vehicles in field areas,and they have important academic significance and application value. |
PDF Full Text Request