基于DenseNet的无人汽车制动意图识别方法

DenseNet-based braking intention recognition method for unmanned vehicles

摘要:

无人汽车制动意图内部数据由于识别深度增加，会出现过度膨胀现象，导致制动意图数据收集完整度低、识别准确率差。提出基于DenseNet的无人汽车制动示意图识别方法。选择数据深度收集系统，收集无人汽车制动意图内部数据，结合电池保护模型深度分解汽车内部运行过程的能耗，以收集的初始内部数据为标准，整合无人汽车制动意图识别数据，拆分整合数据，防止数据过度膨胀。利用DenseNet的高学习度以及自适应学习性，加权均衡处理内部数据标定函数，设置一组基函数，并选择相应的DenseNet复制内部数据函数，自适应分析复制后的数据，完成制动意图识别。实验结果表明，制动意图数据收集完整度提高15.21%，识别准确率增强了23.68%。

Abstract:

Due to the increase of the recognition depth, the inner data of the braking intention of the unmanned vehicle will expand excessively, which leads to the low integrity of the data collection and the poor recognition accuracy. Based on densenet, a recognition method of brake diagram of unmanned vehicle is proposed. Select the deep data collection system, collect the internal data of the unmanned vehicle braking intention, combine the battery protection model to deeply decompose the energy consumption of the internal operation process of the vehicle, integrate the identification data of the unmanned vehicle braking intention based on the initial internal data collected, split and integrate the data to prevent the excessive expansion of the data. Using densenet's high learning degree and self-adaptive learning ability, weighting and equalizing the internal data calibration function, setting a group of basis functions, selecting the corresponding densenet to copy the internal data function, self-adaptive analyzing the copied data, and completing the brake intention recognition. The experimental results show that the integrity of brake intention data collection is improved by 15.21%, and the recognition accuracy is improved by 23.68%.