In response to the concept of energy conservation and environmental protection, a novel composite battery box with BF/PLA composite is proposed. Firstly, the mechanical properties of BF/PLA composite are tested, and it is concluded that the property parameters of BF/PLA composite with 50% BF mass fraction is selected as the material property parameter of subsequent finite element simulation. Subsequently, the statics analysis and constraint modal analysis of the traditional metal battery box are carried out under the typical working conditions of rapid turning and braking under vertical bumping. Based on this, the upper and lower box materials of the battery box except the bracket are replaced by BF/PLA composite. The morphology optimization, topology optimization and free size optimization are carried out with the constraint that the first-order modal vibration frequency is no less than 30 Hz. Compared with the traditional metal battery box, the stiffness and strength of the optimized BF/PLA composite battery box are significantly enhanced. Moreover, the first-order constraint modal frequency increases by 15.5%, and the comprehensive weight reduction ratio reaches 40.88%. Finally, the optimized BF/PLA composite battery box is verified under random vibration, mechanical shock analysis, collision analysis, extrusion and falling ball analysis and drop analysis conditions. Meanwhile, compared with the traditional metal battery box under the same working conditions, the excellent reliability of the composite battery box is highlighted. The proposed BF/PLA composite battery box satisfies the requirements of stiffness and strength performances under various working conditions, which provides theoretical and data support for the application of composite materials in battery box and other automotive components.

To maximize the performance of power batteries and supercapacitors in a hybrid energy storage system (HESS) and to resolve the conflict between the high power demands of electric vehicles and the limitations of high-current charging and discharging of the power battery, a vehicle power demand model incorporating road slope information has been constructed. This paper takes a HESS composed of power battery and supercapacitor as the object, and a rule-based energy management strategy (EMS) based on road slope information is proposed to realize the reasonable distribution and management of energy under the slope condition. According to the slope information of the road ahead, the energy consumption in the next period was predicted, and the supercapacitor is charged and discharged in advance to meet the energy demand of uphill and the energy recovery capacity of downhill to avoid the high current charge and discharge of the battery. Subsequently, the improved EMS performance was simulated under the New York City Cycle (NYCC) driving conditions with additional slope driving conditions. The simulated results indicate that compared to the existing EMS, the proposed EMS based on slope information can effectively distribute the power demand between the power battery and the supercapacitor, can reduce the discharge current and the duration of high-power discharge, and has a 20.4% higher energy recovery efficiency, effectively increasing the cruising range.

To resolve the issues of a deep backbone network, a large model, slow reasoning speed on a mobile terminal, low detection accuracy for small targets and difficulties detecting and recognizing traffic lights in real time and accurately with YOLOv4, a traffic lights recognition method based on improved YOLOv4 is proposed. The lightweight ShuffleNetv2 network is utilized to replace the CSPDarkNet53 network of YOLOv4 to satisfy the requirements of a mobile terminal. The reformed k-means clustering algorithm is applied to generate anchor boxes for avoiding the sensitivity issue of outliers and initial values. A novel attention mechanism named CS²A is added to enhance the extraction capability of effective features. Multiple data augmentation methods are combined to improve the generalization ability of the model. Ultimately, the detection and recognition of traffic lights can be realized. The S²TLD dataset is selected for training and testing, and it can be proved that the recognition accuracy and model size are greatly optimized. Meanwhile, a self-made dataset is selected for training and testing. Compared with the conventional YOLOv4, the recognition accuracy of the proposed algorithm for traffic lights’ state information increases by 1.79%, and the model size decreases by 81.97%. Appropriate scenes are selected for real-vehicle testing and the results demonstrate that the detection speed of the presented algorithm increases by 16%, and the recognition effect for small targets increases by 37% in comparison with conventional YOLOv4.

In order to solve the problems of obstacle avoidance planning and trajectory optimization of unmanned platform in hilly orchards, this paper proposes a dynamic planning algorithm based on Gaussian probability model to predict dynamic obstacle behavior. The technical route and method of theoretical derivation-algorithm design-simulation analysis-real vehicle testing are both adopted. Firstly, the global reference path of orchard is optimized based on the quadratic programming. Secondly, the intentions of surrounding pedestrians and other dynamic obstacles are estimated based on the Gaussian probability model. Subsequently, a dynamic grid map is created based on the Frenet coordinate system. Then, the obstacles avoidance paths are dynamically planned to obtain discrete optimal paths. Finally, the quintic polynomial curve is utilized to connect the sampling points to generate the path with continuous curvature. The analysis by ROS/Rviz simulation indicates that the planning time decreases by 57.97%, the number of sampled nodes decreases by 79.57%, and the curvature is smoother and continuous compared with the RRT algorithm. A real-world test is conducted on a hilly orchard road. With the Gaussian model, the algorithm can effectively predict the behavioral state of dynamic obstacles. The platform can autonomously plan a continuous, smooth and safe driving route in a shorter period of time, effectively avoiding obstacles and reducing the impact of steering-in-place on operating efficiency. Through the research, a set of software and hardware solutions for comprehensive planning of spontaneous static obstacle avoidance in hilly orchards is proposed.