神經網絡在無人駕駛車輛運動控制中的應用綜述

張守武; 王恒; 陳鵬; 張笑語; 李擎

doi:10.13374/j.issn2095-9389.2021.04.23.001

神經網絡在無人駕駛車輛運動控制中的應用綜述

doi: 10.13374/j.issn2095-9389.2021.04.23.001

張守武^{1, 2},
王恒^{1, 3},
陳鵬¹,
張笑語¹,
李擎^{1, 3, ,}

1.
北京科技大學自動化學院，北京100083
2.
北京城市學院信息學部，北京 100083
3.
工業過程知識自動化教育部重點實驗室，北京100083

基金項目: 國家自然科學基金資助項目（61673098，61603034）；北京市自然科學基金資助項目（3182027）；中央高校基本科研業務費資助項目（FRF-GF-17-B44）

詳細信息

通訊作者:
E-mail：liqing@ies.ustb.edu.cn

中圖分類號: TP183
計量
- 文章訪問數: 857
- HTML全文瀏覽量: 371
- PDF下載量: 182
- 被引次數: 0
出版歷程
- 收稿日期: 2021-04-23
- 網絡出版日期: 2021-05-24
- 刊出日期: 2022-02-15

Overview of the application of neural networks in the motion control of unmanned vehicles

ZHANG Shou-wu^{1, 2},
WANG Heng^{1, 3},
CHEN Peng¹,
ZHANG Xiao-yu¹,
LI Qing^{1, 3
, ,}

1.
School of Automation and Electrical Engineering, University of Science and Technology Beijing, Beijing 100083, China
2.
School of Information Science and Engineering, Beijing City University, Beijing 100083, China
3.
Key Laboratory of Knowledge Automation for Industrial Processes, Ministry of Education, Beijing 100083, China

More Information

Corresponding author: E-mail: liqing@ies.ustb.edu.cn

摘要

摘要: 無人駕駛車輛自身具有強烈的非線性、信號時延和參數不確定性，對它的控制還受到道路附著系數的變化、側向風等外界因素影響。因此傳統控制方法往往難以對其穩定和精確地控制。神經網絡所具有的學習能力、自適應能力和近似非線性映射的能力，為解決車輛模型參數的不確定性、外界的擾動以及車輛自適應控制問題提供了有效的途徑。針對上述幾個方面，對近幾年國內外學者將神經網絡應用到無人駕駛車輛運動控制中所取得的成果與進展進行了歸納分類，分別介紹了應用情況并對優缺點進行評價。最后總結了神經網絡在無人駕駛車輛運動控制中存在的主要問題，并展望了可能的發展方向。
- 神經網絡 /
- 非線性系統 /
- 自適應控制 /
- 穩定性 /
- 無人駕駛車輛
Abstract: This paper aims to introduce the application of neural networks in the motion control of unmanned vehicles in recent years. With the breakthrough of computer, robot control, and sensing technology, the development of unmanned vehicles has entered a stage of rapid development. It can reduce driver mistakes, bring convenience to the daily travel of humans, and it is widely used in the military and dangerous fields. However, the unmanned vehicle itself has strong nonlinearity, signal delay, and parameter uncertainty and its control is affected by external factors such as the change of road adhesion coefficient and lateral wind. Therefore, traditional control methods often face challenges in controlling the vehicle stably and accurately. The learning, adaptive, and approximate nonlinear mapping abilities of neural networks provide an effective way to solve the problems of vehicle model parameter uncertainty change, external disturbance, and vehicle adaptive control. Therefore, it is increasingly applied to the motion control of unmanned vehicles. The self-learning and adaptive ability of neural networks enable them to calculate the direct output control quantity according to the state deviation of the vehicle, which can be used as the controller of the unmanned vehicle. The ability of the neural networks to approach a nonlinear mapping makes it possible to approach the unknown dynamic parts of the vehicle, such as the uncertain parameters and external disturbances, which improves the accuracy and robustness of the controller design. The neural networks can remember previous information that can be used to calculate the current output. Thus, the neural networks can be used as the vehicle state observer to estimate the vehicle state parameters. The adaptive ability of the neural networks enables them to optimize the parameters of other control algorithms online. From these aspects, this paper summarized and classified the achievements and progress made by domestic and foreign scholars in applying neural networks to the motion control of unmanned vehicles in recent years, introduced the application situation, and evaluated the advantages and disadvantages. Finally, the main problems of neural networks in the motion control of unmanned vehicles were summarized and the possible development direction was prospected.
- neural network /
- nonlinear system /
- adaptive control /
- stability /
- unmanned vehicle

HTML全文

圖 1 三層前饋神經網絡駕駛員控制器原理圖

Figure 1. Three layer feedforward neural network driver controller schematic diagram

下載: 全尺寸圖片幻燈片

圖 2 基于神經網絡和卡爾曼濾波的汽車側傾角估計原理圖

Figure 2. Vehicle roll Angle estimation schematic based on neural network and Kalman filter

下載: 全尺寸圖片幻燈片

www.77susu.com

參考文獻(56)

[1]	Sun Y, Song L, Wang G Z. A review of the development of foreign ground unmanned autonomous systems in 2019. Aerodyn Missile J, 2020(1): 30 孫毅, 宋樂, 王桂芝. 2019年國外地面無人自主系統發展綜述. 飛航導彈, 2020(1):30
[2]	Li K Q, Dai Y F, Li S B, et al. State-of-the-art and technical trends of intelligent and connected vehicles. J Automot Saf Energy, 2017, 8(01): 1 doi: 10.3969/j.issn.1674-8484.2017.01.001 李克強, 戴一凡, 李升波, 等. 智能網聯汽車(ICV)技術的發展現狀及趨勢. 汽車安全與節能學報, 2017, 8(01):1 doi: 10.3969/j.issn.1674-8484.2017.01.001
[3]	Yang Y M, Gao Z G, Zhang Z L, et al. Research on development strategy of unmanned driving technology. Eng Sci, 2018, 20(6): 101 楊艷明, 高增桂, 張子龍, 等. 無人駕駛技術發展對策研究. 中國工程科學, 2018, 20(6):101
[4]	Bauml T, Giuliani H, Simic D, et al. An advanced simulation tool based on physical modelling of electric drives in automotive applications // 2007 IEEE Vehicle Power and Propulsion Conference. Arlington, 2007: 736
[5]	Cardamone L, Loiacono D, Lanzi P L. On-line neuroevolution applied to The Open Racing Car Simulator // 2009 IEEE Congress on Evolutionary Computation. Trondheim, 2009: 2622
[6]	Hunt K J, Sbarbaro D, ?bikowski R, et al. Neural networks for control systems—A survey. Automatica, 1992, 28(6): 1083 doi: 10.1016/0005-1098(92)90053-I
[7]	Sun F C, Li L, Sun Z Q. Survey on adaptive control of nonlinear systems using neural networks. Control Theory Appl, 2005, 22(2): 254 doi: 10.3969/j.issn.1000-8152.2005.02.016 孫富春, 李莉, 孫增圻. 非線性系統神經網絡自適應控制的發展現狀及展望. 控制理論與應用, 2005, 22(2):254 doi: 10.3969/j.issn.1000-8152.2005.02.016
[8]	Liu J M, Lu X Z, Wang G Q. Application of back propagation neural network to the orientation control of intelligent vehicle. Autom Instrum, 2007(5): 19 doi: 10.3969/j.issn.1001-9227.2007.05.008 柳金梅, 魯學柱, 王國權. BP神經網絡在智能車輛方向控制中的應用. 自動化與儀器儀表, 2007(5):19 doi: 10.3969/j.issn.1001-9227.2007.05.008
[9]	Zhang Y Z, Su H S, Liu G H, et al. Parameter adjustment of PID controller based on BP neural network. Acta Sci Nat Univ Nankaiensis, 2018, 51(3): 26 張永振, 蘇寒松, 劉高華, 等. 基于BP神經網絡的PID控制器參數調整. 南開大學學報(自然科學版), 2018, 51(3):26
[10]	Fei J T, Ding H F. Adaptive sliding mode control of dynamic system using RBF neural network. Nonlinear Dyn, 2012, 70(2): 1563 doi: 10.1007/s11071-012-0556-2
[11]	Xiong Z G, Liu Z, Wang H Y, et al. Design of automatic steering control system based on RBF neural network incremental PID. J Agric Mech Res, 2021, 43(4): 27 doi: 10.3969/j.issn.1003-188X.2021.04.005 熊中剛, 劉忠, 王寒迎, 等. RBF神經網絡增量式PID自動轉向控制系統設計. 農機化研究, 2021, 43(4):27 doi: 10.3969/j.issn.1003-188X.2021.04.005
[12]	Zhang F J, Wang Y, Zhao W Z. Vehicle state estimation of vehicle under limiting conditions based on deep learning. J Chongqing Univ Technol Nat Sci, 2018, 32(10): 64 張鳳嬌, 汪, 趙萬忠. 基于深度學習的極限工況下車輛的狀態估計. 重慶理工大學學報(自然科學), 2018, 32(10):64
[13]	Song R, Fang Y C. Vehicle state estimation for INS/GPS aided by sensors fusion and SCKF-based algorithm. Mech Syst Signal Process, 2021, 150: 107315 doi: 10.1016/j.ymssp.2020.107315
[14]	Pomerleau D A. Neural networks for intelligent vehicles // Proceedings of the Intelligent Vehicle ' 93 Symposium. Tokyo, 1993: 19
[15]	Baluja S. Evolution of an artificial neural network based autonomous land vehicle controller. IEEE Trans Syst Man Cybern B (Cybern), 1996, 26(3): 450 doi: 10.1109/3477.499795
[16]	Demirli K, Khoshnejad M. Autonomous parallel parking of a car-like mobile robot by a neuro-fuzzy sensor-based controller. Fuzzy Sets Syst, 2009, 160(19): 2876 doi: 10.1016/j.fss.2009.01.019
[17]	Zhang W M, Han H B, Yang J, et al. A neural network-based autonomous articulated vehicle system considering driver behavior. J South China Univ Technol Nat Sci, 2016, 44(12): 74 張文明, 韓泓冰, 楊玨, 等. 基于駕駛員行為的神經網絡無人駕駛控制. 華南理工大學學報(自然科學版), 2016, 44(12):74
[18]	Hess R A, Modjtahedzadeh A. A control theoretic model of driver steering behavior. IEEE Control Syst Mag, 1990, 10(5): 3 doi: 10.1109/37.60415
[19]	Guo K H. Preview follower theory and simulations of large angle cornering motion of a man-vehicle system. Automot Eng, 1992, 14(1): 1 郭孔輝. 預瞄跟隨理論與人-車閉環系統大角度操縱運動仿真. 汽車工程, 1992, 14(1):1
[20]	Yu L M, Wang Z L. Optimal preview compensative tracking control for pilot vehicle systems. Acta Automatica, 2001, 27(3): 421 于黎明, 王占林. 人機系統最優預見補償跟蹤控制研究. 自動化學報, 2001, 27(3):421
[21]	LI X Q, HE Y S, XU Z M, et al. Driver Model of Steering Direction Control for Automobiles. J Chongqing Univ Technol Nat Sci, 2006, 29(4): 5 李興泉, 賀巖松, 徐中明, 等. 汽車方向控制駕駛員模型. 重慶大學學報(自然科學版), 2006, 29(4):5
[22]	Ohno H. Analysis and modeling of human driving behaviors using adaptive cruise control. Appl Soft Comput, 2001, 1(3): 237 doi: 10.1016/S1568-4946(01)00022-9
[23]	Guo K H, Pan F, Ma F J. Preview optimized artificial neural network driver model. Chin J Mech Eng, 2003, 39(1): 26 doi: 10.3321/j.issn:0577-6686.2003.01.006 郭孔輝, 潘峰, 馬鳳軍. 預瞄優化神經網絡駕駛員模型. 機械工程學報, 2003, 39(1):26 doi: 10.3321/j.issn:0577-6686.2003.01.006
[24]	Wang H, Zhou F, Li Q, et al. Reliable adaptive H∞ path following control for autonomous ground vehicles in finite frequency domain. J Frankl Inst, 2020, 357(14): 9599 doi: 10.1016/j.jfranklin.2020.07.028
[25]	Gao Y Q, Gray A, Tseng H E, et al. A tube-based robust nonlinear predictive control approach to semiautonomous ground vehicles. Veh Syst Dyn, 2014, 52(6): 802 doi: 10.1080/00423114.2014.902537
[26]	Hang P, Chen X B, Fang S D, et al. Robust control for four-wheel-independent-steering electric vehicle with steer-by-wire system. Int J Automot Technol, 2017, 18(5): 785 doi: 10.1007/s12239-017-0078-5
[27]	Chen T, Chen D. Lateral control of intelligent vehicle based on neural networks sliding mode. Transducer Microsyst Technol, 2017, 36(5): 63 陳濤, 陳東. 基于神經網絡滑模的智能車輛橫向控制. 傳感器與微系統, 2017, 36(5):63
[28]	Kayacan E. Sliding mode control for systems with mismatched time-varying uncertainties via a self-learning disturbance observer. Trans Inst Meas Control, 2019, 41(7): 2039 doi: 10.1177/0142331218794266
[29]	Liu Y H, Li T, Yang Y Y, et al. Estimation of tire-road friction coefficient based on combined APF-IEKF and iteration algorithm. Mech Syst Signal Process, 2017, 88: 25 doi: 10.1016/j.ymssp.2016.07.024
[30]	Hashemi E, Khosravani S, Khajepour A, et al. Longitudinal vehicle state estimation using nonlinear and parameter-varying observers. Mechatronics, 2017, 43: 28 doi: 10.1016/j.mechatronics.2017.02.004
[31]	Zhang S W, Li Q, Wang H, et al. Path Following Control for Autonomous Vehicles with Mismatched Uncertainties. Control Decis, https://doi.org/10.13195/j.kzyjc.2020.1069 張守武, 李擎, 王恒, 等. 非匹配不確定性影響下的無人車路徑跟蹤控制. 控制與決策, https://doi.org/10.13195/j.kzyjc.2020.1069
[32]	Ji X W, He X K, Lv C, et al. Adaptive-neural-network-based robust lateral motion control for autonomous vehicle at driving limits. Control Eng Pract, 2018, 76: 41 doi: 10.1016/j.conengprac.2018.04.007
[33]	Wu Y, Wang L F, Zhang J Z, et al. Path following control of autonomous ground vehicle based on nonsingular terminal sliding mode and active disturbance rejection control. IEEE Trans Veh Technol, 2019, 68(7): 6379 doi: 10.1109/TVT.2019.2916982
[34]	Liu J, Gao L, Zhang J J, et al. Super-twisting algorithm second-order sliding mode control for collision avoidance system based on active front steering and direct yaw moment control. Proc Inst Mech Eng D:J Automob Eng, 2021, 235(1): 43 doi: 10.1177/0954407020948298
[35]	Swain S K, Rath J J, Veluvolu K C. Neural network based robust lateral control for an autonomous vehicle. Electronics, 2021, 10(4): 510 doi: 10.3390/electronics10040510
[36]	Van Zanten A T. Bosch ESP Systems: 5 Years of Experience. SAE Trans, 2000, 109: 428
[37]	Esmailzadeh E, Goodarzi A, Vossoughi G R. Optimal yaw moment control law for improved vehicle handling. Mechatronics, 2003, 13(7): 659 doi: 10.1016/S0957-4158(02)00036-3
[38]	Venhovens P J T, Naab K. Vehicle dynamics estimation using Kalman filters. Veh Syst Dyn, 1999, 32(2-3): 171 doi: 10.1076/vesd.32.2.171.2088
[39]	Cheli F, Sabbioni E, Pesce M, et al. A methodology for vehicle sideslip angle identification: Comparison with experimental data. Veh Syst Dyn, 2007, 45(6): 549 doi: 10.1080/00423110601059112
[40]	Lin F, Zhao Y Q. Comparison of methods for estimating vehicle side slip angle. J Nanjing Univ Sci Technol Nat Sci, 2009, 33(1): 122 林棻, 趙又群. 汽車側偏角估計方法比較. 南京理工大學學報(自然科學版), 2009, 33(1):122
[41]	Wang R R, Jing H, Hu C, et al. Robust $H∞$ path following control for autonomous ground vehicles with delay and data dropout. IEEE Trans Intell Transp Syst, 2016, 17(7): 2042 doi: 10.1109/TITS.2015.2498157
[42]	Wang Y, Lin Y L, Dai X Y, et al. Parallel estimation of vehicle state. J Command Control, 2017, 3(3): 186 doi: 10.3969/j.issn.2096-0204.2017.03.0186 汪, 林懿倫, 戴星原, 等. 車輛關鍵狀態的平行估計. 指揮與控制學報, 2017, 3(3):186 doi: 10.3969/j.issn.2096-0204.2017.03.0186
[43]	Melzi S, Sabbioni E. On the vehicle sideslip angle estimation through neural networks: Numerical and experimental results. Mech Syst Signal Process, 2011, 25(6): 2005 doi: 10.1016/j.ymssp.2010.10.015
[44]	Rajamani R, Piyabongkarn D N. New paradigms for the integration of yaw stability and rollover prevention functions in vehicle stability control. IEEE Trans Intell Transp Syst, 2013, 14(1): 249 doi: 10.1109/TITS.2012.2215856
[45]	Doumiati M , Victorino A, Charara A, et al. Embedded estimation of the tire/road forces and validation in a laboratory vehicle// In Proceedings of 9th International Symposium on Advanced Vehicle Control. Kobe, 2008: 1
[46]	Rajamani R, Piyabongkarn D, Tsourapas V, et al. Parameter and state estimation in vehicle roll dynamics. IEEE Trans Intell Transp Syst, 2011, 12(4): 1558 doi: 10.1109/TITS.2011.2164246
[47]	Jo K, Chu K, Sunwoo M. Interacting multiple model filter-based sensor fusion of GPS with in-vehicle sensors for real-time vehicle positioning. IEEE Trans Intell Transp Syst, 2012, 13(1): 329 doi: 10.1109/TITS.2011.2171033
[48]	Nam K, Oh S, Fujimoto H, et al. Estimation of sideslip and roll angles of electric vehicles using lateral tire force sensors through RLS and Kalman filter approaches. IEEE Trans Ind Electron, 2013, 60(3): 988 doi: 10.1109/TIE.2012.2188874
[49]	Vargas-Meléndez L, Boada B, Boada M, et al. A sensor fusion method based on an integrated neural network and Kalman filter for vehicle roll angle estimation. Sensors, 2016, 16(9): 1400 doi: 10.3390/s16091400
[50]	Boada B L, Boada M J L, Vargas-Melendez L, et al. A robust observer based on H∞ filtering with parameter uncertainties combined with Neural Networks for estimation of vehicle roll angle. Mech Syst Signal Process, 2018, 99: 611 doi: 10.1016/j.ymssp.2017.06.044
[51]	Shao J K, Zhao X, Yang J, et al. Reinforcement learning algorithm for path following control of articulated vehicle. Trans Chin Soc Agric Mach, 2017, 48(3): 376 doi: 10.6041/j.issn.1000-1298.2017.03.048 邵俊愷, 趙翾, 楊玨, 等. 無人駕駛鉸接式車輛強化學習路徑跟蹤控制算法. 農業機械學報, 2017, 48(3):376 doi: 10.6041/j.issn.1000-1298.2017.03.048
[52]	Zhenhai G, Bo Z. Vehicle lane keeping of adaptive PID control with BP neural network self-tuning // 2005 IEEE Proceedings of Intelligent Vehicles Symposium. Las Vegas, 2005: 84
[53]	Yin Z S, He J X, Nie L Z, et al. Longitudinal adaptive control of autonomous vehicles base on optimization algorithm. J Syst Simul, 2021, 33(2): 409 尹智帥, 何嘉雄, 聶琳真, 等. 基于優化算法的自動駕駛車輛縱向自適應控制. 系統仿真學報, 2021, 33(2):409
[54]	Kayacan E, Kayacan E, Ramon H, et al. Towards agrobots: Trajectory control of an autonomous tractor using type-2 fuzzy logic controllers. IEEE/ASME Trans Mechatron, 2015, 20(1): 287 doi: 10.1109/TMECH.2013.2291874
[55]	Xu X, Tang Z, Wang F, et al. Varied weight coefficients based trajectory tracking of distributed drive self-driving vehicle. China J Highw Transp, 2019, 32(12): 36 徐興, 湯趙, 王峰, 等. 基于變權重系數的分布式驅動無人車軌跡跟蹤. 中國公路學報, 2019, 32(12):36
[56]	Zhu F B. Research on Lateral Tracking Control and Mode Switching of Intelligent Vehicle under Typical Conditions [Dissertation]. Zhenjiang: Jiangsu University, 2019 朱方博. 典型工況下的智能車橫向跟蹤控制及模式切換研究[學位論文]. 鎮江: 江蘇大學, 2019