摘要
对于突发紧急威胁情况,常规的无人机规避方法在实时性和适用性方面存在不足.在研究生物体条件反射机制的基础上,将无人机应急规避行为理解为在外界威胁刺激下的一种应激性,提出了基于威胁紧迫度的Skinner理论.模拟飞行员在紧急防撞情况下的拉杆动作,将阶跃信号作为无人机应急动作指令,运用动作评价算法计算输出最佳策略.采用Skinner理论和统计学方法进行在线训练,形成威胁状态与规避动作的匹配,从而建立完整的条件反射过程.实验结果表明,基于Skinner理论的规避方法对突发威胁情况具备有效的规避能力.
The urgent threat collision condition is hazardous for UAV, it is difficult for traditional methods to ensure safety due to the poor performance in real-time and applicability. Conformed to operant conditioning theory, the urgent collision avoidance behavior of UAV could be regarded as irritability in the outside stimulatory, and then the improved Skinner theory based on threat level was proposed. A step command was used as control signal, which was similar with pilot＇s maneuver in the urgent threat collision condition, and performances estimation algorithm was applied to output optimized strategy. On-line training was conducted based on Skinner theory and statistics to map the threat condition and maneuver, then all the elements of operant conditioning model were completed. The result shows that, the proposed method can handle urgent threat collision well.
出处
《北京理工大学学报》
EI
CAS
CSCD
北大核心
2016年第6期620-624,共5页
Transactions of Beijing Institute of Technology
基金
国家自然科学基金资助项目(61573373
61503405)
关键词
无人机
条件反射
威胁规避
应急机制
UAV
operant conditioning
collision avoidance
emergency mechanism
