考虑人机信任匹配的人机协同控制策略

Human-machine Cooperative Control Strategy Considering Human-machine Trust Matching

摘要:

人机互信水平是影响人机协同系统表现的关键因素之一。提出了一种考虑人机信任匹配的主从博弈协同控制策略。建立评估驾驶人和机器相互信任程度的方法，在此基础上根据人机信任匹配程度进行协同驾驶中的权重分配；采用模型预测控制框架并结合主从博弈进行最优化求解，得出最优的协同控制策略；通过驾驶人在环实验，验证了所提出的协同控制策略的有效性。结果表明，对于不同信任匹配程度的驾驶人，所提出的策略使得驾驶人路径跟踪精度平均提高了70.91%，驾驶负担平均降低了44.03%。所提出的策略能提升车辆的驾驶表现，减轻驾驶人操作负担。

Abstract:

The level of human-machine mutual trust is a key factor affecting the performance of human-machine cooperative systems. This paper presents a Stackelberg Game-based cooperative control strategy that considers human-machine trust matching. Firstly， a method was proposed for assessing the mutual trust level between drivers and machines. Based on this， the weight allocation in cooperative driving was performed according to the level of human-machine trust matching. Subsequently， a model predictive control framework was adopted， and the optimal cooperative control strategy was obtained by combining the Stackelberg Game theory for optimization. Finally， driver-in-the-loop experiments were conducted to validate the proposed cooperative control strategy. Results demonstrate that， for drivers with different trust matching levels， the strategy can improve the precision of path tracking by 70.91%， and reduce the driving burden by 44.03%. The proposed strategy enhances the driving performance and reduces the driver workload.