The prime challenge in a humanoid robot is its stability on two feet due to the presence of an underactuated system. In this paper, the complete dynamics of the humanoid robot has been described in essence of torque calculation at the end effectors. Presence of various restraints in humanoid robot motion makes the task of stabilization an even humongous one. Therefore, to neutralize these constraints, whole-body control (WBC) has been proposed to consider the free-floating base and to ensure the stability of the humanoid robot. Dynamic modeling of the humanoid robot is performed based on the Langrage–Euler formalism to obtain the maximum torque at the joints. This approach is utilized to formulate the torque equation and solve the problem of stabilization. WBC deals with the limitation of attainment of well nimble dynamics behavior operated at high speeds. The simulated annealing approach is preferred to tune WBC to get efficient stabilization and eliminate the earlier limitation. In addition, the zero-moment point (ZMP) criterion is taken care of as it affects the stability of the humanoid robot aggressively. Simulations on V-REP are carried out to understand the torque behavior at each joint. To validate the simulation results, the experiments are carried out on the NAO humanoid robot in real experimental conditions. The experimental and simulation results are compared through torque versus time graphs, and they both show good agreement with deviation under 4% between them. The proposed technique is then compared with various previously implemented techniques which confirm the robustness and efficiency of the proposed methodology.

中文翻译：

---

基于模拟退火的全身控制NAO仿人机器人动态稳定

人形机器人的主要挑战是由于存在欠驱动系统，它在两只脚上的稳定性。本文从末端执行器扭矩计算的本质描述了仿人机器人的完整动力学。人形机器人运动中存在各种限制，使得稳定任务变得更加艰巨。因此，为了消除这些限制，提出了全身控制（WBC）来考虑自由浮动底座并确保仿人机器人的稳定性。基于 Langrage-Euler 形式对仿人机器人进行动态建模，以获得关节处的最大扭矩。该方法用于制定扭矩方程并解决稳定问题。WBC 处理在高速运行时获得良好灵活的动力学行为的限制。首选模拟退火方法来调整 WBC 以获得有效的稳定性并消除早期的限制。此外，考虑了零矩点（ZMP）标准，因为它会严重影响仿人机器人的稳定性。对 V-REP 进行了模拟，以了解每个关节处的扭矩行为。为了验证仿真结果，在真实实验条件下对 NAO 仿人机器人进行了实验。实验和仿真结果通过扭矩与时间曲线图进行比较，两者均显示出良好的一致性，偏差在 4% 以下。