Scholarlink

<p>The main objective of the paper is to address the challenges of secure navigation for a biped robot in uncertain and unknown environments. Smooth trajectories are generated to mitigate collision risks and potential damage from obstacles and ditches, whose positions and shapes are not known in prior. Two crucial aspects considered here are path planning and trajectory planning for the biped’s walk. To achieve this, the proposed methodology utilizes Q learning for optimal path planning followed by the generation of efficient trajectories using FNN and WNN under several constraints. Problem of generating smooth trajectories by adapting the real-time changes in the environment and adjusting constraint values online accordingly is considered. A 5-degree-of-freedom (DOF) biped robot is considered for illustrating the results, evaluating its ability to adjust its gait/step during tracking, generated by using FNN and WNN according to the Q learning based path in uncertain environment. Simulation experiments in Matlab2014a validate the smoothness of trajectories and their suitability for the biped robot’s walk on uneven terrains. Additional criteria including Zero Moment Point (ZMP) stability and PD controller are included in the tracking control design.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Biped robot’s locomotion in complex environment using different neural networks and Q learning

  • 期刊论文

  • 发表日期：

    2025年12月12日

查看全文
  • Ruchi Panwar,
  • N. Sukavanam

摘要

The main objective of the paper is to address the challenges of secure navigation for a biped robot in uncertain and unknown environments. Smooth trajectories are generated to mitigate collision risks and potential damage from obstacles and ditches, whose positions and shapes are not known in prior. Two crucial aspects considered here are path planning and trajectory planning for the biped’s walk. To achieve this, the proposed methodology utilizes Q learning for optimal path planning followed by the generation of efficient trajectories using FNN and WNN under several constraints. Problem of generating smooth trajectories by adapting the real-time changes in the environment and adjusting constraint values online accordingly is considered. A 5-degree-of-freedom (DOF) biped robot is considered for illustrating the results, evaluating its ability to adjust its gait/step during tracking, generated by using FNN and WNN according to the Q learning based path in uncertain environment. Simulation experiments in Matlab2014a validate the smoothness of trajectories and their suitability for the biped robot’s walk on uneven terrains. Additional criteria including Zero Moment Point (ZMP) stability and PD controller are included in the tracking control design.