【作者】 王猛；

【导师】 蔡鹤皋； 赵杰；

【作者基本信息】 哈尔滨工业大学 ， 机械电子工程， 2009， 博士

【摘要】 仿生跳跃机器人研究是机器人领域的一个前沿性课题。由于机器人不仅具备生物体结构和行为方式合理、灵活和高效的特征,其跳跃运动方式还能适应不同地表,实现跨越沟渠和障碍,具有活动范围广、躲避风险能力强的特点,展现了优异的移动能力,因此适合在非结构化、不可预测的环境里代替人类完成侦察、探测、救险和反恐等任务,有着重要的理论研究意义和广阔的应用前景。青蛙跳跃具有爆发性强、距离远的特点,拥有这种能力的跳跃机器人将能轻松越过障碍,并且具有很好的环境适应性。因此本文选择以青蛙为研究对象,进行仿青蛙跳跃机器人的研究。青蛙跳跃运动的轨迹和姿态变化规律是研究仿生机器人生物运动机理的基础,针对其运动信息难以获得的问题,提出了一种空间运动轨迹提取方法,在进行透视效果和迭代修正后,可得到关节点的三维坐标。针对逐点逐帧分析造成计算量大的问题,建立了一套采集程序,最终可输出青蛙各点位置、速度和加速度曲线及其各个关节角随时间的变化曲线。在对该方法的正确性进行验证后,进行了青蛙跳跃轨迹提取实验,获得了青蛙质心和各关节运动信息。然后基于对青蛙跳跃运动的观察和获取的运动信息,对青蛙的解剖学生物特征,跳跃运动特征和关节轨迹进行分析,得出了后肢关节轨迹变化的相似性、起跳阶段脚掌跗跖关节的重要功能以及前后肢在跳跃过程中各自的作用。通过动力学仿真研究了其跳跃运动过程中的力学规律,所得到的分析结果将用于仿青蛙跳跃机器人的设计和控制。基于对青蛙生物特征和跳跃运动机理的分析,对复杂的关节、骨骼结构进行简化,提出了一种面向跳跃运动的机构模型,并进行仿青蛙跳跃机器人设计。机器人前肢简化为一个主动肩关节和一个被动肘关节,从而实现其着陆支撑缓冲和姿态调整的功能。后肢采用五杆机构作为腿部主体,并增加脚掌和被动跗跖关节以保证跳跃和着陆时的稳定性,髋关节具有后摆和外摆两个自由度以调整跳跃的姿态、方向和轨迹。机器人可以调整多种起跳姿态以满足不同的跳跃要求;其合理的质量分布有利于提高能量转化效率;后肢通过单电机分时控制实现跳跃、空中收腿和能量调节动作,具有较高的功率密度;后肢五杆机构具有与青蛙跳跃时相似的力学规律,起跳时通过与脚掌配合,可以降低提前离开地面的可能,从而降低了能量损耗。运动学和动力学分析是机器人姿态调整、跳跃轨迹规划和实时控制的基础。对机器人准备和跳跃的各个阶段进行了正、逆运动学分析,建立了运动学方程,研究了腾空阶段机器人前后肢末端的运动空间。针对机器人前后肢结构复杂、不利于动力学分析的问题,基于虚功原理对其行了简化,建立了前肢和后肢运动模型,然后对机器人起跳、腾空和着陆三个阶段以及离地和触地两个瞬时进行了动力学分析,建立了起跳阶段后肢动力学方程、腾空阶段机器人运动方程、着陆阶段前肢动力学方程、离地瞬时起跳判据和触地瞬时翻转判据,并通过仿真验证了运动学和动力学方程的正确性。机器人的跳跃轨迹规划是使其运动满足指标要求的重要手段,是跳跃机器人完整路径规划的基础。针对动力学方程多变量和非线性造成轨迹规划算法复杂的问题,首先对机器人跳跃轨迹进行分析,然后提出了基于遗传算法的机器人跳跃轨迹多参数多目标规划方法,建立了机器人跃过各种典型地形应满足的约束条件,以跳跃稳定性、安全性、能量消耗和水平跳跃距离为目标函数,对机器人跨越水平沟渠的跳跃进行了轨迹规划实验,通过动力学计算和仿真验证了该方法的有效性。最后,采用分级和模块化思想,设计了机器人上位机进行运动规划、嵌入式控制器进行关节控制的控制系统,然后对机器腿单元跳跃、机器人姿态调整和机器人水平跳跃做了实验研究,并且将机器人跳跃与青蛙跳跃过程进行对比分析,实验结果表明了以仿生学方法提取出的机构模型的有效性、仿青蛙跳跃机器人结构设计的合理性和理论分析的正确性。更多还原

【Abstract】 The research on biomimetic jumping robot is a forward positional task in robot field. The robot owns the feature of reasonable, agile and high-efficiency in structure and movement like biosome, whose jumping movement could adjust to different ground, realize getting across cannal and obstacle. The robot reveals excellent ability on movement territory and avoiding risk, so it is suitable to achieve the task such as spying, detecting, rescue, anti-terror and so on in non-frame and indeterminism circumstance instead of human. It is important for theory research and has wide application prospect. Frog’s jumping movement has the characteristic of strong burst and long distance, the robot that owns this ability could cross obstacle easily and has good adaptability for circumstance. Therefore, this paper selects frog as study target to do research on frog-like jumping robot.Considering that the trajectory and the stance of frog jumping movement are the basis of the research on the biological jumping mechanism of biomimetic robot design, a new method of trajectory extraction from animal’s movement is presented. And the 3D coordinate of the joint is obtained after perspective effect and iteration correction. To overcome the shortcoming of large amount of calculation, a program is designed to deal with the video images. And finally the curve of position, velocity and acceleration of each point and the angle of each joint of frog are exported. After the validation of this method, the frog movement information of center of mass and joint is obtained by experiment. Based on the observation and obtained movement information, the jumping movement characteristics and the joint trajectory are analyzed. we find the similar trajectories of hindlimb joints during jump, the important effect of foot during take-off phase and the role of forelimb and hindlimb during jump phase. The force during jump is stuied by dynamic simulation. The analysis result will be used for design and control of frog jumping robot.Based on the analysis of biological characteristics and jumping movement mechanism, frog’s complex joint and skeleton structure is simplified. A mechanical model is put forward and the frog jumping robot is designed. The robot’s forelimb is simplified as one active shoulder joint and one passive elbow joint for the function of supporting and stance adjusting. The 5-bar lingkage mechanism is used as the main part of the hindlimb, the foot and passive tarsometa-tarsal joint are added for guaranteing the stability of jumping and landing. The two DOFs of coxa joint are used for adjusting the pose, direction and trajectory of jump. The robot could achieve multi-stance to satisfy different demand for jump. The reasonable distribution of quality is helpful for improving efficiency. The hindlimb could realize the movement of jump, leg constriction in air and energy adjusting by single motion, which owns high density of actuator. The 5-bar linkage mechanism of hindlimb cooperated with foot has similar process of the thrust force with frog during take-off phase, which possesses the advantages of low likehood of premature lift-off and low energy loss.The kinematic and dynamic analysis is the base of robot stance adjusting, jumping trajectory planning and real-time control. After direct and inverse kinematic analysis on each phase of robot movement, the kinematic equation is established and the movement space of limb’s terminal during aerial phase is studied. The complex forelimb and hindlimb of robot are simplified with virtual work principle for dynamics analysis. Based on the movement model of four limbs, the hindlimb dynamics equation during take-off pahse, the robot kinematics equation during aerial phase, the forelimb dynamics equation during landing pahse, the take-off judgement at off-ground instantaneous and the eversion judgement at touchdown instantaneous are eatablished through the dynamic analysis of whole jump process of robot. The kinematic and dynamic equations are verified by simulation.The jumping trajectory planning is the important method to make robot movement satisfied with the target demand. It is also the base of path planning for jumping robot. Considering that non-line dynamics equation with multi-variable make trajectory planning algorithm to be complex, we first analyze the jumping trajectory, then bring forward the trajectory planning method for multi-parameter and multi-target based on genetic algorithm. The conditions for crossing typical terrain are established. Under the target function of jump stability, security, energy consuming and horizontal distance, the trajectroy planning experiment about crossing canal is done. The method’s availability is verified by dynamic computing and simulation.Finally, the control system is desigened based on hierarchical and modularization strategy, which includes upper computer for movement trajectory planning and embedded controller for joint control. Then the experiment about the robot leg, stance adjusting and horizontal jump is implemented, and the jump process is compared with frog. The experiment result proves that the mechanism modle extracted with biomimetic method is valid, and the mechanical design of the frog jumping robot is reasonable and the theory analysis is accurate.更多还原