Legged robots with enhanced terrain adaptability

Mobile robots can assist humans in performing exploration and rescue tasks in dangerous environments and have broad application prospects in fields such as mountain transportation, firefighting and rescue, geological exploration, and military operations. Mobile robots are mainly categorized into wheeled robots, tracked robots, and biomimetic legged robots. 事行动等领域有广阔的应用前景。移动机器人主要分为轮式机器人、履带式机器人和仿生足式机器人。

Among them, legged robots have stronger terrain adaptability and motion stability compared to other types; they can move at high speeds on flat ground, as well as adapt to complex terrains such as hills, grassland, and uneven sandy surfaces.

Bionic leg robots are divided into bipedal, quadrupedal, hexapodal, and other multi-legged robots. Unlike bipedal robots, which balance on a single foot during walking, quadrupedal bionic robots have at least two points of contact with the ground, achieving near-complete balance when all four feet are on the ground. Therefore, quadrupedal and hexapodal robots are more capable of balancing control on complex terrains. Due to their better stability and load-bearing capacity, multi-legged robots are likely to be integrated into our lives sooner.

The research and development of biomimetic multi-legged robots is a comprehensive and widely applicable scientific project, involving the analysis of animal biomimetics, kinematics, dynamics, mechanical design of robots, gait planning, motion simulation, and control among multiple disciplines. By studying the body structure and functional mechanisms of quadrupedal organisms and utilizing theories of kinematics and dynamics, the driving mechanisms and mechanical structures of biomimetic quadruped robots can be determined. Based on the motion process of quadrupedal animals, the gait of biomimetic quadruped robots can be planned periodically. Signals received by different sensors are processed by the control system, which then provides feedback control to the servo motors at the joints of the biomimetic quadruped robot, maintaining stability during walking. Gait planning and control are core elements in the design process of the biomimetic quadruped robot and are prerequisites for the practical use of multi-legged robots.

To improve the efficiency and off-road capabilities of mechanical walking, developers use correlation analysis methods based on animals like goats and cheetahs, which have excellent stability and displacement abilities in unstructured environments. They analyze the coordination between the animal joints and apply their movement mechanisms and behavioral patterns to the study of legged robot walking. Motion capture technology is used to capture the reflective markers at key points on the animals' bodies during movement, acquiring 3D spatial coordinates of the markers. After processing, various parameters of the animal's limb movements are obtained, including step frequency, stride, walking speed, step length, step width, support phase, swing phase, joint angles, and angular velocities. These parameters are used to study the kinematic characteristics of animal movement, such as the temporal patterns of different gaits and the changes in joint displacement and joint angular motion.

Additionally, by using a 3D motion capture system combined with the robot's own sensor information, the robot's dynamic parameters can be acquired in real-time for analysis and gait planning. The planned gait is then transmitted to the robot's control system, which controls the servo motors to drive the robot's movement and corrects posture in real-time. Utilizing the NOKOV optical 3D motion capture system, the absolute positions of key points on the robot's body can be captured in the world coordinate system. By using the motion capture system's SDK for real-time broadcasting, feedback is provided for the robot to analyze and process. The system can also extract marker point coordinate information offline, obtaining the robot's position and posture, and validating the robustness of the robot and control algorithms.

Researchers both domestically and internationally are increasingly developing multi-legged robots. Among these, the products with higher maturity levels are the multi-legged robots from Boston Dynamics. The Little Dog robot developed by Boston Dynamics is a typical representative of quadruped robots. Each leg of the Little Dog robot contains 3 degrees of freedom, with joint movements driven by servo motors. To focus research on gait control and improve the robot's ability to overcome rough terrain, researchers use the robot's own equipment to identify features of its surrounding 3D environment and construct an environmental map. They have also set up a motion capture system composed of multiple cameras to determine the robot's absolute position and orientation in the world coordinate system using markers placed on the robot.

The implementation of an intelligent agent cluster experimental system requires indoor simultaneous localization of multiple agents, and due to the small indoor space, high localization accuracy is demanded.

Shanghai Jiao Tong University Hexapod Robot Project

NOKOV工程师在长9米宽6米的空间内，架设了8个Mars 2HMotion Capture Cameras，以60Hz的Sampling Frequency捕捉机器人“躯干”和“四肢”关节上的反光标志点，得到各标志点三维坐标，确定相互间位置，从而获取动态运动学数据。动作捕捉完毕后，利用Seeker软件对捕捉数据进行三维重建通过动作捕捉系统自带的 SDK把这些数据实时广播，机器人团队进行实时的分析和反馈，得以确认六足机器人各足的姿态，实现六足的校准标定以及各腿间的动作协调。

Shandong University Quadruped Bionic Robot Project

At a national earthquake emergency response training base's test field, NOKOV engineers set up 12 NOKOV Mars 2H cameras on tripods over an outdoor area of 14 meters in length and 8 meters in width. Reflective markers were attached to the joints of the "trunk" and "limbs" of the biomimetic robot. The motion of the robot was captured at a sampling rate of 60Hz, and the spatial coordinates of each joint were exported with sub-millimeter precision. Users then imported these data into specialized analysis tools for analysis to confirm the robot's posture and develop gait improvement plans.