四足机器人校园巡逻自适应视觉方法研究

摘 要：随着人工智能专业以及机器人技术的快速发展，人工智能专业创新创业实训将四足机器人校园巡逻方法纳入到课程教学体系当中。四足机器人凭借优异的地形适应能力与运动灵活性，成为校园巡逻等半结构化场景巡检的重要载体。本文以四足机器人校园巡逻为应用场景，阐述基于自适应视觉感知的巡检系统从方案论证到落地测试的全流程。本系统以 Unitree Robotics Go2 EDU 四足机器人作为研究对象，按照视觉层与控制层分开设计的两层解耦方案搭建起“感知—决策—执行”的完整软硬结合的层次式整体框架，实现了基于HSV颜色空间的自适应寻线、棕色标志点指令化变向以及红色台阶区域爬阶模式转换，以及绿色障碍物视觉避障等功能。实验结果表明，在典型的校园仿真场景下本系统的各个功能模块识别准确率达到80%以上，全流程的任务完成率超过90%。实现了算法研究-电路开发-程序编写-实验测试的一整套流程，并锻炼了多学科交叉合作、软硬件联合设计以及系统的反复修改完善的能力，在校企协同育人模式下的机器人专业教学中可以作为范例。

关键词：具身智能；HSV色彩空间；感知决策执行架构；自适应视觉感知；四足机器人

正文 

可下载并阅读全文PDF，请按照本文版权许可使用。
Download the full text PDF for viewing and using it according to the license of this paper.

Research on Adaptive Visual Method for Quadruped Robot Campus Patrol

Abstract: With the rapid development of artificial intelligence majors and robotics technology, the innovation and entrepreneurship training of artificial intelligence majors has incorporated the quadruped robot campus patrol method into the curriculum teaching system. With excellent terrain adaptability and movement flexibility, quadruped robots have become an important carrier for patrolling semi-structured scenes such as campus patrols. This article takes the four-leged robot campus patrol application scenario to explain the whole process of the inspection system based on adaptive visual perception from scheme demonstration to landing test. This system takes the UnitreeRobotics Go2 EDU quadruped robot as the research object, and builds a hierarchical overall framework of "perception, decision-making, execution" complete soft and hard combination according to the two-layer decoupling scheme designed separately from the visual layer and the control layer, and realizes the HSV-based Adaptive line search of color space, direction change of brown sign point instruction, step mode conversion in red step area, and visual obstacle avoidance of green obstacles and other functions. The experimental results show that under the typical campus simulation scenario, the identification accuracy of each functional module of the system reaches more than 80%, and the task completion rate of the whole process exceeds 90%. It has realized a whole set of processes of algorithm research-circuit development-program writing-experimental testing, and exercised the ability of multidisciplinary cross-cooperation, joint design of software and hardware, and repeated modification and improvement of the system. It can be used as an example in the professional teaching of robots in the mode of collaborative education between schools and enterprises.

Keywords: Embodied Intelligence; HSV Color Space; Perception-Decision-Execution Architecture; Adaptive Visual Perception; Quadruped Robot

参考文献

1. 陈晓东,梁永康,连宗凯.防爆四足机器人在成品油油库的自主巡检系统设计与安全评估[J].产品可靠性报告,2025,(10):37-38.
2. 付鑫,朱西硕.煤矿四足机器人巡检系统设计与研发[J].智能矿山.2025,(11):71-75..
3. 王堃,唐清琼,付环.面向LNG接收站的四足机器人巡检系统设计与研究[J].电器工业.2025,(8):71-75.
4. 贾琪,李艳,徐洋,等.基于中枢预测控制的四足机器人多地形运动控制[J].东华大学学报(自然科学版),2026,(7):1-16.
5. 国家互联网信息办公室. “十四五” 数字经济发展规划 [Z].2022.
6. Sridharan M, Stone P. Color learning and illumination invariance on mobile robots: A survey[J]. Robotics and Autonomous Systems, 2009,57(6-7):629-644.
7. Mur artal Raúl, Montiel J.M.M, Tardós Juan D. ORB SLAM: A versatile and accurate monocular SLAM system[J]. IEEE Transactions on Robotics, 2015, 31(5): 1147-1163.
8. 同爱丽.实时多任务调度方法研究与应用 [D]. 西安：西北工业大学，2006.
9. Lee Joonho, Bjelonic Marko, Reske Alexander, et al. Learning robust autonomous navigation and locomotion for wheeled-legged robots[J]. Science Robotics,2024,9:eadi9641.
10. 孙帅帅,冯春晓,张良.基于离散采样的多模态四足机器人路径规划[J].吉林大学学报:工学版.2025,(11):3736-3744.
11. 章敏晋.图像工程(上册):图像获取与预处理[M].北京:清华大学出版社,2012.
12. Wozniak Piotr, Penar Paweł, Bielecki Damian. RGB-Based Staircase Detection for Quadrupedal Robots: Implementation and Analysis[J]. Sensors, 2025, 25(23): 7247.
13. 曾明海,李澍源.基于相位指导的四足机器人步态切换方法研究[J].机械设计与制造,2026,(7):286-291.
14. O. Khatib. Real-time obstacle avoidance for manipulators and mobile robots[C] //Proceedings of 1985 IEEE International Conference on Robotics and Automation. St. Louis: IEEE,1985:500-505.
15. 王康.四足机器人复杂地形越障控制方法研究[D].北京:北京交通大学,2023.
16. 修春良,范永,刘大宇,等.四足机器人动静态路径规划方法研究[J].现代信息科技,2025,9(24):192-198.
17. 杨婉霖,贾豫东.四足机器人颜色识别算法设计[J].计算机科学与应用, 2024, 14(12):111-117.
18. 董夏男,蒋学程.URDF与Simulink结合的四足机器人仿真研究[J].机械管理开发,2025,40(08):272-275.