摘要
为了提升农业智能车辆在动态环境中换道及应对障碍轨迹变化的能力,构建混合整数随机滚动优化决策框架,通过高斯过程对主车与障碍车未来轨迹分布进行非参数建模;结合蒙特卡罗模拟与分离轴定理评估动态碰撞风险,设计高低风险双优化模型,并采用并行化求解架构与改进遗传算法实现实时求解。所提框架依据碰撞风险阈值动态切换优化模式,高风险场景以最小化碰撞概率为目标,低风险场景综合考虑安全性、效率与舒适性。实验结果表明,该方法平均碰撞概率为0.08±0.02,显著低于传统确定性模型的0.25±0.08;在直线换道路段平均换道时间为3.2±0.4 s,较传统方法提升28.9%;纵向加速度均方根为0.35±0.08 m/s^(2),低于传统方法。该研究可为智能农机在复杂田间环境中的应用提供可靠技术方案。
To enhance the ability of agricultural intelligent vehicles to change lanes and respond to variations in obstacle trajectories in dynamic environments,a mixed-integer random rolling optimization decision framework was constructed.Non-parametric modeling of the future trajectory distribution of the main vehicle and obstacle vehicles was carried out through Gaussian processes,and dynamic collision risk was evaluated by combining Monte Carlo simulation and off-axis theorem.A high and low-risk dual optimization model was designed,and a parallel solution architecture and improved genetic algorithm were used to achieve real-time solution.The proposed framework dynamically switches optimization modes based on collision risk thresholds,with the goal of minimizing collision probability in high-risk scenarios and considering safety,efficiency,and comfort comprehensively in low-risk scenarios.Experimental results showed that the average collision probability of the proposed method is 0.08±0.02,significantly lower than that of the traditional deterministic model(0.25±0.08);the average time for changing lanes in straight sections is 3.2±0.4 s,which is 28.9%longer than that of traditional methods;the root mean square of longitudinal acceleration is 0.35±0.08 m/s^(2),which is lower than taht of traditional methods.This study provides a reliable technical solution for the application of intelligent agricultural machinery in complex field environments.
作者
于春鹏
范德会
YU Chunpeng;FAN Dehui(School of Automobile and Traffic Engineering,Heilongjiang Institute of Technology,Harbin 150050,China)
出处
《国外电子测量技术》
2025年第5期212-217,共6页
Foreign Electronic Measurement Technology
关键词
农业智能车辆
换道优化
障碍轨迹
实时优化
agricultural intelligent vehicles
lane change optimization
obstacle trajectory
real-time optimization
