Scholarlink

In the field of robotics, ensuring real-time control responses is an imperative for high-performance applications. Conventional iterative methods for calculating inverse kinematics are computationally intensive and tend to introduce significant delays in the control cycle of robots, thus compromising operational efficiency. This research presents an innovative application of a multilayer perceptron (MLP) neural network, carefully designed to accelerate inverse kinematics calculations far beyond the capabilities of conventional approaches. The network architecture has been optimized to improve calculation speed while maintaining accuracy. Specifically, the results reveal that this MLP-based approach reduces calculation times by up to 150 times compared to traditional iterative solutions while maintaining positional accuracy. These results are very promising for working in real time with robotic systems.

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

Enhancing Robotic Control Efficiency with MLP-Based Inverse Kinematics: First Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • M. Peñacoba-Yagüe,
  • J. E. Sierra-García,
  • M. Santos-Peñas,
  • A. Ruano

摘要

In the field of robotics, ensuring real-time control responses is an imperative for high-performance applications. Conventional iterative methods for calculating inverse kinematics are computationally intensive and tend to introduce significant delays in the control cycle of robots, thus compromising operational efficiency. This research presents an innovative application of a multilayer perceptron (MLP) neural network, carefully designed to accelerate inverse kinematics calculations far beyond the capabilities of conventional approaches. The network architecture has been optimized to improve calculation speed while maintaining accuracy. Specifically, the results reveal that this MLP-based approach reduces calculation times by up to 150 times compared to traditional iterative solutions while maintaining positional accuracy. These results are very promising for working in real time with robotic systems.