四足机器人特定复杂运动技能控制

doi:10.12382/bgxb.2023.0874

摘要/Abstract

摘要：

为了提高四足机器人的运动多样性和地形适应性,提出一种复杂运动行为控制方法,通过构建四足机器人动力学模型,在此基础上进行离线滚动优化预测,生成四足机器人复杂行为的期望轨迹。在运动学、关节扭矩、接触力、运动状态和地形高度等非线性约束下更全面地优化了轨迹,设计在线轨迹跟踪控制器与落足控制器,实现四足机器人复杂行为的在线控制。在多复杂运动的动态仿真环境下评估了所提方法,机器人可以实现前跳、后空翻、前空翻和旋转跳跃,并可以在给定地形信息下跳跃障碍物。将在线轨迹跟踪控制器迁移到四足机器人物理样机中,完成了四足机器人向前跳跃的实验。实验结果表明,所提出的方法能够使四足机器人有效地完成多种特定复杂运动技能的稳定控制。

关键词: 四足机器人, 复杂行为, 离线滚动优化预测, 在线轨迹跟踪控制器

Abstract:

A complex locomotion behavior control method is proposedto improve the locomotion diversity and terrain adaptability of quadruped robot. A dynamics model for the quadruped robots is established, and then the offline rolling optimization is predicted to generate the desired trajectory of robot’s complex locomotion behavior. The locomotiontrajectoriesof robot under more comprehensive nonlinear constrains, such as kinematics, joint torque, contact force, locomotion state, and terrain height,etc, are optimized. An online trajectory tracker and a foot placement hopping controller are designed to realize the online control of the quadruped robot. The proposed method is evaluated in dynamic simulation environment of multi complex locomotion. The robot can achieve front jump, backflip, forward flip and rotary jump, and can also jump over obstacles according to the given terrain information. Finally, the online trajectory tracking controller is migrated to the physical prototype of the quadruped robot, and the forward jumping test of the quadruped robot is completed. The experimental results show that the proposed method can be used effectively to achieve the stable control of various specific complex locomotion motion skills for the quadruped robot.

Key words: quadruped robot, complex behavior, off-line rolling optimization prediction, on-line trajectory tracking controller

中图分类号:

TP242.3

许鹏, 赵建新, 范文慧, 邱天奇, 江磊, 梁振杰, 刘宇飞. 四足机器人特定复杂运动技能控制[J]. 兵工学报, 2023, 44(S2): 135-145.

XU Peng, ZHAO Jianxin, FAN Wenhui, QIU Tianqi, JIANG Lei, LIANG Zhenjie, LIU Yufei. Specific Complex Locomotion Skills Control for Quadruped Robots[J]. Acta Armamentarii, 2023, 44(S2): 135-145.

图/表 15

图1 四足机器人模型

Fig.1 Quadruped robot model

图2 四足机器人动力学模型

Fig.2 Dynamics model of quadruped robot

图3 四足机器人复杂行为控制框图

Fig.3 Block diagramof complex behavior controlof quadruped robot

图4 四足机器人后空翻序列

Fig.4 Backflip sequence of quadruped robot

图5 四足机器人前空翻序列

Fig.5 Forward flip sequence of quadruped robot

图6 四足机器人向前跳序列

Fig.6 Forward jump sequence of quadruped robot

图7 四足机器人旋转跳跃序列

Fig.7 Rotary jump sequence of quadruped robot

图8 四足机器人后空翻数据

Fig.8 Backflip data of quadruped robot

图9 四足机器人前空翻数据

Fig.9 Forward flip data of quadruped robot

图10 四足机器人向前跳数据

Fig.10 Forward jump data of quadruped robot

图11 四足机器人旋转跳数据

Fig.11 Rotary jump data of quadruped robot

图12 四足机器人跳跃障碍序列

Fig.12 Jumpobstacles sequence of quadruped robot

图13 四足机器人跳跃障碍物数据

Fig.13 Jumpobstacles data of quadruped robot

图14 物理样机向前跳序列

Fig.14 Forward jump sequenceof physical prototype

图15 物理样机向前跳数据

Fig.15 Forward jump data of physical prototype

参考文献 26

[1]	Boston Dynamics. Introducing Spot (previouslySpotMini[EB/OL]. (2016-06-23)[2023-12-03]. https://www.youtube.com/watch?v=tf7IEVTDjng.
[2]	Boston Dynamics. Cheetah robot runs 28.3 mph; a bit faster than Usain Bolt[EB/OL]. (2012-09-06) [2023-12-03]. https://www.youtube.com/watch?v=chPanW0QWhA&t=7s.
[3]	Boston Dynamics. LS3-Legged Squad Support System[EB/OL]. (2012-09-11)[2023-12-03]. https://www.youtube.com/watch?v=R7ezXBEBE6U.
[4]	PARK H W, WENSING P M, KIM S. Online planning for autonomous running jumps over obstacles in high-speed quadrupeds[C]// Proceedings of the 11st Robotics Science and Systems. Rome, Italy: MIT, 2015:1-9.
[5]	HYUN D J, SEOK S, LEE J, et al. High speed trot-running: Implementation of a hierarchical controller using proprioceptive impedance control on the MIT Cheetah[J]. The International Journal of Robotics Research, 2015, 33(11):1417-1445. doi: 10.1177/0278364914532150 URL
[6]	PARK H W, CHUAH M Y, KIM S, et al. Quadruped bounding control with variable duty cycle via vertical impulse scaling[C]// Proceedings of the 26th Intelligent Robots and Systems. Chicago, IL, US: IEEE, 2014:3245-3252.
[7]	DI C J, WENSING P M, KATZ B, et al. Dynamic locomotion in the mit cheetah 3 through convex model-predictive control[C]// Proceedings of the 30th Intelligent Robots and Systems. Madrid, Spain:IEEE, 2018: 1-9.
[8]	KIM D, DI C J, KATZ B, et al. Highly dynamic quadruped locomotion via whole-body impulse control and model predictive control[DB/OL]. (2019-09-14) [2023-12-03]. https://arxiv.org/abs/2110.02799.
[9]	GEHRING C, COROS S, HUTTER M, et al. Control of dynamic gaits for a quadrupedal robot[C]// Proceedings of the 29th International Conference on Robotics and Automation. Karlsruhe, Germany: IEEE, 2013:3287-3292.
[10]	GEHRING C, BELLICOSO C D, COROS S, et al. Dynamic trotting on slopes for quadrupedal robots[C]// Proceedings of the 27th Intelligent Robots and Systems. Hamburg, Germany: IEEE, 2015:5129-5135.
[11]	BJELONIC M, GRANDIA R, GEILINGER M, et al. Offline motion libraries and online MPC for advanced mobility skills[J]. The International Journal of Robotics Research, 2022, 41(9/10): 903-924. doi: 10.1177/02783649221102473 URL
[12]	MIKI T, LEE J, HWANGBO J, et al. Learning robust perceptive locomotion for quadrupedal robots in the wild[J]. Science Robotics, 2022, 7(62): eabk2822. doi: 10.1126/scirobotics.abk2822 URL
[13]	HWANGBO J, LEE J, DOSOVITSKIY A, et al. Learning agile and dynamic motor skills for legged robots[J]. Science Robotics, 2019, 4(26): eaau5872. doi: 10.1126/scirobotics.aau5872 URL
[14]	LEE J, HWANGBO J, WELLHAUSEN L, et al. Learning quadrupedal locomotion over challenging terrain[J]. Science Robotics, 2020, 5(47): eabc5986. doi: 10.1126/scirobotics.abc5986 URL
[15]	梁启星, 李彬, 李志, 等. 基于模型预测控制的四足机器人斜坡自适应调整算法与实现[J]. 山东大学学报 (工学版), 2021, 51(3): 37-44.
	LIANG Q X, LI B, LI Z, et al. Algorithm of adaptive slope adjustment of quadruped robot based on model predictive control and its application[J]. Journal of Shandong University(Engineering Science), 2021, 51(3): 37-44. (in Chinese)
[16]	HAN K C, KIM J Y. Posture stabilizing control of quadruped robot based on cart-inverted pendulum model[J]. Intelligent Service Robotics, 2023, 16:521-536. doi: 10.1007/s11370-023-00480-8
[17]	SOMBOLESTAN M, NGUYEN Q. Hierarchical adaptive loco-manipulation control for quadruped robots[C]// Proceedings of the 39th International Conference on Robotics and Automation.London, UK: IEEE, 2023:12156-12162.
[18]	张国腾, 荣学文, 李贻斌, 等. 基于虚拟模型的四足机器人对角小跑步态控制方法[J]. 机器人, 2016, 38(1): 64-74. doi: 10.13973/j.cnki.robot.2016.0064
	ZHANG G T, RONG X W, LI Y B, et al. Control of the quadrupedal trotting based on virtual model[J]. Robot, 2016, 38(1): 64-74.(in Chuese) doi: 10.13973/j.cnki.robot.2016.0064
[19]	ZHAO J X, YAO Q C, XING B Y, et al. One-legged hop of compliance control based on minimum-jerk[J]. Journal of Physics Conference Series, 2020, 1507(5): 052012. doi: 10.1088/1742-6596/1507/5/052012
[20]	孟健, 李贻斌, 李彬. 四足机器人跳跃步态控制方法[J]. 山东大学学报(工学版), 2015, 45(3):28-34.
	MENG J, LI Y B, LI B, et al. Bound gait controlling method of quadruped robot[J]. Journal of Shandong University (Engineering Science), 2015, 45(3):28-34. (in Chinese)
[21]	PARK H W, PARK S, KIM S. Variable-speed quadrupedal bounding using impulse planning:Untethered high-speed 3D Running of MIT Cheetah 2[C]// Proceedings of the 31st International Conference on Robotics and Automation. Seattle, WA, US: IEEE, 2015: 5163-5170.
[22]	BJELONIC M. Planning and control for hybrid locomotion of wheeled-legged robots[D]. ETH Zurich, 2021.
[23]	HOELLER D, RUDIN N, SAKO D, et al. Learning agile navigation for quadrupedal robots[DB/OL]. (2023-06-26) [2023-12-03]. https://arxiv.org/abs/2306.14874.
[24]	HAN L, ZHU Q, SHENG J, et al. Lifelike agility and play on quadrupedal robots using reinforcement learning and generative pre-trained models[DB/OL]. (2023-08-29) [2023-12-03]. https://arxiv.org/abs/2308.15143.
[25]	NGUYEN Q, POWELL M J, KATZ B, et al. Optimized jumping on the mit cheetah 3 robot[C]// Proceedings of the 35th International Conference on Robotics and Automation. Montreal, Canada: IEEE, 2019: 7448-7454.
[26]	CHIGNOLI M M T. Trajectory optimization for dynamic aerial motions of legged robots[D]. Cambridge, MA, US:MIT, 2021.