自由落猫机器人最优下落轨迹研究

发布时间：2018-03-31 22:35

本文选题：落猫现象　切入点：空中姿态调整　出处：《中国科学技术大学》2017年硕士论文

【摘要】：猫从高处坠落能安全落地的能力一直被研究者们关注。当一只猫从高空跌落时,它总能通过姿态调整使四肢朝下,加上腿部具有减少地面冲击作用,最终能实现安全落地。机器人在未知复杂环境下作业时,有从高处跌落的风险,所以需考虑其空中的姿态调整能力,减轻由错误的着陆方式造成的伤害。因此研究落猫的空中翻正原理,将它的这些身体特点和运动特征应用到机器人上可期待解决机器人高空坠落的安全着陆问题。本文主要研究落猫机器人在自由下落过程中的最优下落轨迹问题。首先把落猫机器人简化为由两个轴对称刚体组成的双刚体系统,以其前后半身能否扭转作为判别标准,把机器人模型分为不可扭转模型与可扭转模型,分别建立了动力学方程。由于受到非完整约束的约束,落猫机器人系统是一个非完整系统,该系统轨迹规划问题可转化为非完整运动规划问题。选择机器人各关节耗能作为最优控制目标函数,通过控制输入参数化方法把无限维最优化问题转化为有限维最优化问题,运用罚函数方法把有约束最优化问题转化为无约束优化问题,分别采用拟牛顿算法和粒子群优化算法求出目标函数的最优解。考虑到机器人做自由落体运动时,一般下落高度固定,提供给其在空中进行姿态调整的时间非常有限,因此相对于能量最优化方法来讲,缩短空中姿态调整时间要比减少能量消耗更加重要且具有实际意义,随后首次提出了以时间最优方式研究落猫机器人最优下落轨迹问题。最后制作了虚拟样机和实物样机,并分别在虚拟物理环境和垂直悬挂条件下,对样机进行了验证实验,结果显示,两种样机都能实现空中翻正动作,进而验证了本文所提方法的有效性。本文最后对所取得的研究成果进行了总结,指出了目前研究中存在的不足之处,并对下一步的研究方向提供了自己的看法。
[Abstract]:The ability of a cat to fall safely from a height has been the focus of researchers. When a cat falls from a high altitude, it always adjusts its posture to lower its limbs, and the legs have the ability to reduce ground impact. The robot has the risk of falling from a height when it is operating in an unknown and complex environment, so it needs to consider its ability to adjust its attitude in the air. To reduce the damage caused by the wrong landing. The application of these body and motion features to the robot can be expected to solve the problem of safe landing of the robot falling from high altitude. This paper mainly studies the optimal falling trajectory of the cat-dropping robot in the process of free falling. First of all, the cat-dropping robot is simplified as a two-rigid-body system consisting of two axisymmetric rigid bodies. Based on the criterion of torsion in front and rear body, the robot model is divided into irreversible model and reversible model, and the dynamic equations are established respectively. Because of the constraint of nonholonomic constraint, the dynamic equations of the robot are established. Cat falling robot system is a nonholonomic system. The trajectory planning problem of the system can be transformed into a nonholonomic motion planning problem. The energy consumption of each joint of the robot is chosen as the optimal control objective function. By using the control input parameterization method, the infinite dimensional optimization problem is transformed into a finite dimensional optimization problem, and the penalty function method is used to transform the constrained optimization problem into an unconstrained optimization problem. Quasi Newton algorithm and particle swarm optimization algorithm are used to find the optimal solution of the objective function respectively. Considering that when the robot moves in free fall, the general falling height is fixed, and the time for attitude adjustment in the air is very limited. Therefore, compared with the energy optimization method, it is more important and practical to shorten the time of attitude adjustment than to reduce the energy consumption. Finally, the virtual prototype and the physical prototype are made, and the prototype is verified in the virtual physical environment and the vertical suspension condition, respectively. The results show that both of the two prototypes can realize the aerial righting motion, which verifies the effectiveness of the method proposed in this paper. Finally, the research results obtained are summarized, and the shortcomings of the present research are pointed out. And the next research direction to provide their own views.
【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】