Omnidirectional mobile robots encounter uncertain nonlinear factors during operation that substantially affect tracking performance. This paper proposes a hybrid HGAPSO algorithm with parallel population partitioning and bidirectional information exchange to optimize PID controller parameters for four-wheeled omnidirectional robots. Unlike sequential hybrid approaches, HGAPSO maintains simultaneous exploration (GA) and exploitation (PSO) capabilities throughout optimization. Comprehensive experimental validation on a physical platform demonstrates superior performance: trajectory accuracy improved 38% vs GA-PID (0.029 → 0.018 m) and 22% vs PSO-PID (0.023 → 0.018 m); settling time reduced 48–51%, overshoot suppressed 53–67%. Robustness tests under payload variations (±100%), external disturbances (30 N), surface friction changes, and sensor noise confirm stable performance with 35–45% less degradation than baseline methods.