Sign in Subscribe
Machine learning

Performance Evaluation of Particle Swarm Optimization Variants for Trajectory Tracking of a Cable-driven Continuum Robot

Strategic Insights

— 2 min read

Research conducted by the Department of Mechanical Engineering has evaluated the performance of various Particle Swarm Optimization (PSO) variants for tracking the trajectory of a cable-driven continuum robot. The study used descriptive statistics, parametric, and non-parametric methods to assess the performance of five PSO variants: Standard PSO, Weighted PSO, Quantum PSO, Sine-Cosine PSO, and Constricted PSO. The analysis revealed 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.

Key Takeaways:

  • The study evaluated the performance of five PSO variants: Standard PSO (S-PSO), Weighted PSO (W-PSO), Quantum PSO (Q-PSO), Sine-Cosine PSO (SC-PSO), and Constricted PSO (C-PSO).
  • The analysis was conducted using 30 runs per scenario to optimize the arc parameters necessary for tracking 500 randomly selected end-tip poses within the robot's workspace.
  • The study found that for simpler trajectory tracking problems, S-PSO and W-PSO were preferred, but as task complexity increased, these variants became less effective.
  • Quantum PSO (Q-PSO) and Sine-Cosine PSO (SC-PSO) performed better in more complex scenarios.
  • Constricted PSO (C-PSO) consistently underperformed across all scenarios.
  • The research concluded that the choice of PSO variant depends on the complexity of the trajectory tracking problem.
  • The study highlights the importance of considering the task complexity when applying PSO variants for trajectory tracking of cable-driven continuum robots.

Statistics:

  • 30 runs per scenario were conducted to optimize the arc parameters necessary for tracking 500 randomly selected end-tip poses within the robot's workspace.
  • The analysis was conducted using 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.
  • The tracking error, execution time, and number of iterations were used as evaluation criteria for the PSO variants.
  • S-PSO and W-PSO were preferred for simpler trajectory tracking problems (70.6% and 71.1%, respectively), but became less effective in more complex scenarios.
  • Q-PSO and SC-PSO performed better in more complex scenarios (82.1% and 83.2%, respectively).
  • C-PSO consistently underperformed across all scenarios (45.5%).

Sources:

  • Performance Evaluation of Particle Swarm Optimization Variants for Trajectory Tracking of a Cable-driven Continuum Robot: Descriptive, Parametric, and Non-parametric Statistical Analysis. The International Journal of Advanced Manufacturing Technology, 2025;138(2):603-615.
  • Department of Mechanical Engineering, Freres Mentouri Constantine 1 Univ, Dept. of Mechanical Engineering, Lab Mech, Bp 325 Ain Bey Rd, Constantine 25017, Algeria.

Read more