# Bio-Inspired Metaheuristics for Time-Optimal Trajectory Planning in Cooperative Dual-Arm Bimanipulation

## TL;DR

This paper introduces a new method using bio-inspired algorithms to plan efficient and collision-free movements for dual-arm robots in industrial settings.

## Contribution

The paper introduces a feasibility-first cost structuring and benchmarks PSO, WOA, and GOA for dual-arm trajectory planning.

## Key findings

- PSO achieves the shortest execution time (6.825 s) among collision-free trajectories.
- All three algorithms reliably find feasible cooperative trajectories.
- PSO outperforms WOA and GOA in both feasibility discovery and final trajectory quality.

## Abstract

This paper addresses the generation of time-efficient, collision-free cooperative motions for a dual-arm robotic system transporting a shared payload in constrained industrial workspaces. Trajectory generation is formulated as a constrained optimization problem and solved through bio-inspired metaheuristic search, where candidate solutions are evaluated with a safety-first cost function that first enforces feasibility by heavily penalizing collisions and then minimizes total execution time among collision-free trajectories. Particle Swarm Optimization (PSO), Whale Optimization Algorithm (WOA), and Gazelle Optimization Algorithm (GOA) are evaluated under identical bounds and stopping conditions, showing that all three reliably discover feasible cooperative trajectories; however, clear differences emerge in feasibility discovery and final trajectory quality: PSO reaches feasibility earlier and achieves the lowest final objective value and the shortest trajectory execution time (6.825 s), followed by WOA (7.330 s) and GOA (8.525 s). Overall, this work contributes an object-centric optimization methodology for constrained dual-arm bimanipulation using bio-inspired metaheuristics, a feasibility-first cost structuring that explicitly separates safe motion discovery from time-optimal refinement, and a controlled benchmarking of PSO/WOA/GOA that quantifies their distinct convergence and late-stage refinement behaviors.