Scholarlink

<p>This paper presents a comprehensive evaluation of Particle Swarm Optimization (PSO) variants for trajectory tracking of a cable-driven continuum robot, utilizing descriptive statistical metrics along with parametric and non-parametric methods for performance assessment. The forward kinematic model of the robot was derived using the constant curvature approach, and the trajectory tracking problem was formulated as an optimization task. Five PSO variants, namely Standard PSO (S-PSO), Weighted PSO (W-PSO), Quantum PSO (Q-PSO), Sine-Cosine PSO (SC-PSO), and Constricted PSO (C-PSO), were reviewed and applied to two optimization scenarios: achieving a target point without considering end-tip orientation as a basic optimization problem and achieving both position and orientation as a more complex optimization problem. To evaluate their effectiveness and robustness, each algorithm was run 30 times per scenario to optimize the arc parameters necessary for tracking 500 randomly selected end-tip poses within the robot’s workspace. Descriptive statistics, as well as parametric and non-parametric statistical tests, including one-way ANOVA, Kruskal-Wallis, and Dunn’s post hoc test with Holm-Bonferroni correction, were used to compare the PSO variants based on tracking error, execution time, and number of iterations. The analysis showed that for simpler trajectory tracking problems, S-PSO and W-PSO were preferred, but as task complexity increased, these variants became less effective, with Q-PSO and SC-PSO performing better in more complex scenarios. Meanwhile, C-PSO consistently underperformed across all scenarios.</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

Performance evaluation of Particle Swarm Optimization variants for trajectory tracking of a cable-driven continuum robot: descriptive, parametric, and non-parametric statistical analysis

  • 期刊论文

  • 发表日期：

    2025年4月22日

查看全文
  • Ammar Amouri,
  • Yazid Laib Dit Leksir,
  • Halim Merabti,
  • Ayman Belkhiri,
  • Abdelhakim Cherfia

摘要

This paper presents a comprehensive evaluation of Particle Swarm Optimization (PSO) variants for trajectory tracking of a cable-driven continuum robot, utilizing descriptive statistical metrics along with parametric and non-parametric methods for performance assessment. The forward kinematic model of the robot was derived using the constant curvature approach, and the trajectory tracking problem was formulated as an optimization task. Five PSO variants, namely Standard PSO (S-PSO), Weighted PSO (W-PSO), Quantum PSO (Q-PSO), Sine-Cosine PSO (SC-PSO), and Constricted PSO (C-PSO), were reviewed and applied to two optimization scenarios: achieving a target point without considering end-tip orientation as a basic optimization problem and achieving both position and orientation as a more complex optimization problem. To evaluate their effectiveness and robustness, each algorithm was run 30 times per scenario to optimize the arc parameters necessary for tracking 500 randomly selected end-tip poses within the robot’s workspace. Descriptive statistics, as well as parametric and non-parametric statistical tests, including one-way ANOVA, Kruskal-Wallis, and Dunn’s post hoc test with Holm-Bonferroni correction, were used to compare the PSO variants based on tracking error, execution time, and number of iterations. The analysis showed that for simpler trajectory tracking problems, S-PSO and W-PSO were preferred, but as task complexity increased, these variants became less effective, with Q-PSO and SC-PSO performing better in more complex scenarios. Meanwhile, C-PSO consistently underperformed across all scenarios.