基于物理-数据驱动的轮腿运输机器人运动控制*

doi:10.16731/j.cnki.1671-3133.2026.04.005

现代制造工程 ›› 2026, Vol. 547 ›› Issue (4): 37-44.doi: 10.16731/j.cnki.1671-3133.2026.04.005

基于物理-数据驱动的轮腿运输机器人运动控制^*

李杰^1,2, 郑允昊^1,2, 郦骏楠^1,2, 宋伟森^1,2

1 北京建筑大学机电与车辆工程学院,北京 100044;
2 北京建筑大学城市轨道交通车辆服役性能保障北京市重点实验室,北京 100044

收稿日期:2025-03-04 发布日期:2026-05-07
通讯作者: 郑允昊,硕士研究生,主要研究方向为机器人控制与路径规划、无人车辆智能控制。E-mail:zhengyh0922@qq.com
作者简介:李杰,教授,博士,主要研究方向为机器人控制与路径规划、无人车辆智能控制、车辆传动。E-mail:lijie7721@126.com
基金资助:
^*国家自然科学基金项目(51675494);北京建筑大学金字塔人才培养工程项目(JDJQ20200308)

Motion control of wheeled-legged transport robots based on physics-data driven approach

LI Jie^1,2, ZHENG Yunhao^1,2, LI Junnan^1,2, SONG Weisen^1,2

1 School of Mechanical,Electrical and Vehicle Engineering,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;
2 Beijing Key Laboratory of Performance Guarantee on Urban Rail Transit Vehicles,Beijing University of Civil Engineering and Architecture,Beijing 100044,China

Received:2025-03-04 Published:2026-05-07

摘要/Abstract

摘要： 运输易燃固体粉末或特殊液体时,由于其流动性,被运输物资会发生摩擦生热、静电积聚及自发分解等现象,可能引发火情或爆炸等危险情况,故运输工具需要具备稳定的姿态保持能力,同时具有一定的越野性能。基于此,设计一种新型串联式六轮腿运输机器人,并基于分层控制理论,构建一种拥有腿部扭矩控制器、轮胎转速控制器和TensorFlow腿部角度控制器的控制框架。腿部扭矩控制器基于虚拟模型控制(Virtual Model Control,VMC)算法控制机器人侧倾角、俯仰角和Z方向位移,使其拥有较高的姿态保持能力;轮胎转速控制器利用差速轮运动模型控制横摆角、X和Y方向位移,使其有灵活的转弯性能并减少横向力冲击;引入TensorFlow神经网络框架,依据不同高度的障碍物来控制腿部各个关节角度。对模型进行联合仿真,以验证斜坡工况和台阶工况的效果。结果表明,控制算法能使机器人在斜坡工况中有优秀的姿态保持能力,位置跟踪偏差率均小于5 %,姿态角的偏差率均小于3 %,相比未引入神经网络的情况,引入神经网络后,机器人可翻越障碍物的高度大幅提升,同时在越障过程中的稳定性更高。

关键词: 轮腿运输机器人, 虚拟模型控制, 差速轮运动模型, PID控制, 姿态控制, TensorFlow神经网络

Abstract: When transporting flammable solid powders or special liquids,due to the fluidity,the transported materials may experience effects such as heat generation from friction,electrostatic accumulation,and spontaneous decomposition,which may trigger dangerous situations such as fire or explosion. Therefore,it is necessary for the transportation vehicle to possess excellent attitude maintenance capability and certain off-road performance. Based on this,a new type of series-connected six-wheeled-legged transport robot is designed. And based on the hierarchical control theory,a control framework is constructed,which includes a leg torque controller,a tire rotation speed controller,and a TensorFlow leg angle controller. The leg torque controller employs the Virtual Model Control (VMC) algorithm to control the roll angle,pitch angle,and displacement in the Z direction of the robot,enabling it to have a high attitude maintenance capability. The tire rotation speed controller uses the differential wheel motion model to control the yaw angle,as well as displacements in X and Y directions,endowing it with flexible turning performance and reducing lateral force impacts. The TensorFlow neural network framework is introduced to control the angles of each joint of the legs according to obstacles of different heights. Finally,a co-simulation of the model is carried out to verify the effects under the slope working condition and the step working condition. The results show that the control system can enable the robot to have excellent attitude maintenance capability under the slope working condition,with a position tracking deviation rate less than 5 % and an attitude angle deviation rate being less than 3 %. Compared with the situation without the introduction of the neural network,after the introduction of the neural network,the height of the obstacles that the robot can overcome is significantly increased,and at the same time,the stability during the obstacle crossing process is higher.

Key words: wheeled-legged transport robot, Virtual Model Control (VMC), differential wheel motion model, PID control, attitude control, TensorFlow neural network

中图分类号:

TP24

李杰, 郑允昊, 郦骏楠, 宋伟森. 基于物理-数据驱动的轮腿运输机器人运动控制^*[J]. 现代制造工程, 2026, 547(4): 37-44.

LI Jie, ZHENG Yunhao, LI Junnan, SONG Weisen. Motion control of wheeled-legged transport robots based on physics-data driven approach[J]. Modern Manufacturing Engineering, 2026, 547(4): 37-44.

参考文献

[1] 陈潇凯,王泓晔,刘宏宇,等.智能底盘半主动悬架与地面切向力协调控制研究[J].机械工程学报,2025,61(14):233-242.
[2] 李靖玮,罗建南,黄震.考虑CDC系统时滞的半主动悬架控制器设计研究[J].汽车工程,2024,46(5):913-922.
[3] KATSUYAM E,YAMAKADO M,ABE M. A state-of-the-art review:toward a novel vehicle dynamics control concept taking the driveline of electric vehicles into account as promising control actuators[J]. Vehicle System Dynamics,2021,59(7):976-1025.
[4] KLAMT T,SCHWARZ M,LENZ C,et al. Remote mobile manipulation with the centauro robot:full-body telepresence and autonomous operator assistance[J]. Journal of Field Robotics,2020,37(5):889-919.
[5] KASHIRI N,BACCELLIERE L,MURATORE L,et al. Centauro:a hybrid locomotion and high power resilient manipulation platform[J]. IEEE Robotics and Automation Letters,2019,4(2):1595-1602.
[6] KAMEDULA M,KASHIRI N,TSAGARAKIS N G. Wheeled motion kinematics and control of a hybrid mobility centauro robot[J]. Robotics and Autonomous Systems,2020,128:103482.
[7] MEDEIROS V S,JELAVIC E,BJELONIC M,et al. Trajectory optimization for wheeled-legged quadrupedal robots driving in challenging terrain[J]. IEEE Robotics and Automation Letters,2020,5(3):4172-4179.
[8] 刘尚非,王军政,刘冬琛,等.多功能并联式六轮足救援机器人[J].机械工程学报,2024,60(21):14-26.
[9] 秦日鹏,徐坤,陈佳伟,等.一种星球探测六足轮腿机器人的设计与运动规划[J].航空学报,2021,42(1):154-164.
[10] NI L,MA F,GE L,et al. Design and posture control of a wheel-legged robot with actively passively transformable suspension system[J]. Journal of Mechanisms and Robo-tics,2021,13(1):011014.
[11] WANG S,CUI L,ZHANG J,et al. Balance control of a novel wheel-legged robot:design and experiments[C]//2021 IEEE International Conference on Robotics and Automation (ICRA).[S.l.]:IEEE,2021:6782-6788.
[12] 姜祎,王挺,邵沛瑶,等.一种轮腿复合型机器人的步态研究与越障性能分析[J].兵工学报,2023,44(1):247-259.
[13] 张延军,韩雨.基于改进DWA的四轮差速移动底盘模型算法研究[J].组合机床与自动化加工技术,2024(9):98-103,107.
[14] UMAM F,ADIPUTRA F,DAFID A,et al. Autonomous museum tour guide robot with object detection using tensorflow learning machine[C]//2022 IEEE 8th Information Technology International Seminar (ITIS).[S.l.]:IEEE,2022:274-281.
[15] 王小涛,张震,崔宇新,等.月面六足机器人机构设计与行走控制方法综述[J].载人航天,2023,29(2):264-275.
[16] 崔宇新,韩亮亮,谷程鹏,等.一种适应月面极端地形六支链轮足机器人设计与步态验证[J].华中科技大学学报(自然科学版),2024,52(8):21-28.

基于物理-数据驱动的轮腿运输机器人运动控制^*

Motion control of wheeled-legged transport robots based on physics-data driven approach

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

[1]	张梦琰, 郝玉鹏, 杨胜强, 赵畦延, 李秀红. 铸造机匣弯管型腔的振动抛磨姿态控制与工艺优化研究^*[J]. 现代制造工程, 2026, 547(4): 70-77.
[2]	营江澎, 孙有平, 吴光庆, 陈鹏宇, 季宏春. 电子机械制动夹紧力闭环控制策略研究^*[J]. 现代制造工程, 2025, 538(7): 48-56.
[3]	胡凤宝, 戴士杰, 季文彬, 刘淑媛. 基于MRAC+PID算法的机器人打磨力跟踪控制研究^*[J]. 现代制造工程, 2024, 523(4): 57-65.
[4]	张立军，马哲，王钰文，张晓东，贺庆强，王智伟，王吉岱，马龙，王晓强，李明. 高压输电线路机器人防滑挂线轮毂的H型3D打印装置的研究[J]. 现代制造工程, 2023, 513(6): 20-25.
[5]	邓召文,孔昕昕,高伟. 重型半挂汽车列车主动转向控制策略研究[J]. 现代制造工程, 2021, 493(10): 17-24.
[6]	张保家,杨涛,张有信,王凌鹏. 基于模糊PID的微细金属丝拉拔张力控制研究[J]. 现代制造工程, 2021, 493(10): 107-113.
[7]	胡鸿志,黄俊锋,管芳,覃畅,苏海涛. 基于专家PID的钻削优化控制方法研究[J]. 现代制造工程, 2021, 493(10): 114-118.
[8]	拜颖乾，高攀科，任锐. 盾构机推进液压系统仿真控制研究[J]. 现代制造工程, 2020, 480(9): 128-135.
[9]	孙磊厚，朱旭阳，逸铭，刘光新，王前进. 电缸综合性能测试系统设计[J]. 现代制造工程, 2020, 476(5): 145-149.
[10]	曹竞豪，黄欢，张帅，周颖文. 基于最小方差自校正的整平机姿态控制[J]. 现代制造工程, 2020, 475(4): 6-12.
[11]	李鑫军，柳江，刘之涵，付延轩. 馈能悬架BP神经网络PID控制研究[J]. 现代制造工程, 2020, 474(3): 60-65.
[12]	刘权文，王晓笋，谢福起，巫世晶. 基于AFSA优化的龙门起重机防摆PID控制器[J]. 现代制造工程, 2020, 472(1): 143-148.
[13]	詹小辉, 陈新元, 邓江洪, 俞志鹏, 革军. 加渣机器人渣厚模糊控制系统设计[J]. 现代制造工程, 2019, 460(1): 55-58.
[14]	张崇俊, 阎春平. 通井机下放作业专家模糊PID控制技术研究及应用[J]. 现代制造工程, 2018, 452(5): 112-117.
[15]	李海青, 文强. 基于模糊-PID控制策略的越野车辆主动悬架联合仿真研究[J]. 现代制造工程, 2018, 451(4): 99-102.