为了提高尾桨卡滞下无人直升机轨迹规划的实时性,提出了一种基于模仿学习的飞行轨迹规划方法,使用非线性模型预测控制(Model Predictive Control,MPC)轨迹规划器求解最优着陆轨迹,并将其作为专家策略收集不同情景下的专家示教行为数据以构建模仿学习数据库,进一步搭建深度神经网络进行行为克隆,并采用数据聚合方法提高深度神经网络性能,进而对专家策略进行模仿学习。提出的方法在不同复杂仿真场景下均能规划合理轨迹,并且行为克隆网络相较于非线性MPC轨迹规划器规划时间更短,说明提出的方法具有更好的实时性。
To improve the real-time performance of trajectory planning for unmanned helicopters with tail rotor jamming, a flight trajectory planning method based on imitation learning is proposed. A nonlinear Model Predictive Control (MPC) trajectory planner is used to solve the optimal landing trajectory and is treated as the expert strategy. This planner is used to collect expert demonstration data in various scenarios to construct an imitation learning database. Subsequently, a deep neural network is built to perform behavior cloning, with data aggregation methods applied to enhance the network's performance, thereby enabling the imitation of the expert strategy. The proposed method can plan reasonable trajectories in various complex simulation scenarios, and the behavior cloning network has a shorter planning time compared to the nonlinear MPC trajectory planner, indicating better real-time performance.
[1] 张健, 阮文强. 直升机特情事故调查及分析[J].民航学报, 2018,2(5):46-49.
ZHANG J, RUAN W Q. Investigation and analysis of helicopter special circumstances accidents[J]. Journal of Civil Aviation, 2018,2(5):46-49.
[2] 孙杰, 高艳辉. 直升机尾桨故障及其试飞研究[J].飞行力学, 2001,19(4):67-69.
SUN J, GAO Y H. The helicopter tail-rotor failure and the flight test[J]. Flight Dynamics, 2001,19(4):67-69.
[3] 费景荣, 张朋. 直升机尾桨卡滞后着陆的操控分析[J]. 直升机技术, 2022, 214(4):50-53,58.
FEI J R, ZHANG P. Control analysis on landing of helicopter tail rotor seized[J]. Helicopter Technique, 2022,214(4): 50-53,58.
[4] 严旭飞, 陈仁良, 辛冀. 直升机不同尾桨距卡滞后的着陆轨迹和操纵优化[J]. 西北工业大学学报, 2019,37(6):1138-1147.
YAN X F, CHEN R L, XIN J. Helicopter landing trajectory optimization after tail rotor control failure in different collective pitch[J]. Journal of Northwestern Polytechnical University, 2019,37(6):1138-1147.
[5] 曲成威, 刘天林, 林惟凯, 等. 机器人学习方法综述[J]. 北京大学学报(自然科学版), 2023,59(6):1069-1086.
QU C W, LIU T L, LIN W K, et al. A review of robot learning[J]. Acta Scientiarum Naturalium Universitatis Pekinensis, 2023, 59(6): 1069-1086.
[6] ABBEEL P, COATES A, NG A Y. Autonomous helicopter aerobatics through apprenticeship learning[J]. The International Journal of Robotics Research, 2010,29(13):1608-1639.
[7] 宋紫阳, 李军怀, 王怀军, 等. 基于路径模仿和SAC强化学习的机械臂路径规划算法[J]. 计算机应用, 2024,44(2):439-444.
SONG Z Y, LI J H, WANG H, et al. Path planning algorithm of manipulator based on path imitation and SAC reinforcement learning[J]. Journal of Computer Application, 2024,44(2):439-444.
[8] ZUCKER M, RATLIFF N, STOLLE M, Optimization and learning for rough terrain legged locomotion[J]. The International Journal of Robotics Research, 2011,30(2):175-191.
[9] WANG W L, CHEN R L. Study of helicopter autorotation landing following engine failure based on a six-degree-of-freedom rigid-body dynamic model[J]. Chinese Journal of Aeronautics, 2013,57(2): 1380-1388.
[10] 鲜斌, 宋宁. 基于模型预测控制与改进人工势场法的多无人机路径规划[J]. 控制与决策, 2024,39(7):2133-2141.
XIAN B, SONG N. A multiple UAV path planning method based on model predictive control and improved artificial potential field[J].Control and Decision, 2024,39(7):2133-2141.
[11] DALAMAGKIDIS K, VALAVANIS K P, PIEGL L A. Autonomous autorotation of unmanned rotorcraft using nonlinear model predictive control[J]. Journal of Intelligent & Robotic Systems, 2010,57(2):351-369.
[12] SAKAWA Y. Trajectory planning of a free-flying robot by using the optimal control[J]. Optimal Control Application & Methods, 1999,20(5):235-248.
[13] EBERLE B E, ROGERS J D. Nonlinear model predictive control of a helicopter in autorotative flare[J]. Journal of the American Helicopter Society, 2010,57(2): 351-369.
[14] HARATI E. Nonlinear model predictive controller toolbox[DB/OL].[2024-01-15].https://publications.lib.chalmers.se/records/fulltext/146434.pdf.
[15] 何信弟, 赵超轮, 戴邵武, 等. 威胁环境下四旋翼的轨迹规划与跟踪控制研究[J]. 兵器装备工程学报, 2023,44(11):270-278.
HE X D, ZHAO C L, DAI S W, et al. Research on trajectory planning and tracking control of quadrotor in threat environment[J]. Journal of Ordnance Equipment Engineering, 2023,44(11): 270-278.
[16] GUENTER F, HERSCH M, CALINON S, et al. Reinforcement learning for imitating constrained reaching movements[J]. Advanced Robotics, 2007, 21(13): 1521-1544.
[17] TAKAYUKI O, JONI P, GERHARD N, et al. An algorithmic perspective on imitation learning[J]. Foundations and Trends in Robotics, 2018, 7(1/2): 1-179.
[18] ROSS S, BAGNELL J A. Reinforcement and imitation learning via interactive no-regret learning[J]. Eprint Arxiv, 2014, 25(6):1315-1326.
