<p>The field of vision-guided wheeled mobile robots (WMRs) often encounters complex nonlinear dynamics, varying environmental conditions, and stringent positioning demands. The domain of vision-guided mobile robots (MRs) frequently confronts intricate nonlinear dynamics, variable environmental conditions, and stringent positioning requirements. This necessitates the enhancement of advanced nonlinear approximations to effectively control and observe these autonomous mobile systems. The research paper addresses the significant challenge of enabling autonomous navigation for four-wheeled omnidirectional mobile robots (FWOMRs), with an emphasis on the development of intelligent algorithms for optimal path identification. To accommodate fluctuating lighting conditions, the paper proposes a multimodal object detection framework, YOLO3D-CSSA, designed to enhance the safety and efficiency of FWOMR’s path planning. The cross-scale attention (CSSA) modules facilitate effective information exchange, thereby reducing misalignment and discrepancies between the dual RGB-IR channels within the network backbone. Leveraging the outputs of YOLO3D-CSSA, two-dimensional object bounding boxes (2D-OBB) are utilized within regression models to estimate the dimensions of three-dimensional object bounding boxes (3D-OBB) in FWOMR operational settings. Subsequently, the estimated centers of the 3D-OBBs inform a genetic algorithm (GA), which is rigorously optimized to develop an obstacle avoidance strategy based on a bird’s eye view (BEV) perspective. Furthermore, a hybrid HGAPSO-PID controller has been developed for FWOMRs to accurately track desired trajectories. This hybrid algorithm combines two optimization techniques: particle swarm optimization (PSO) and genetic algorithms (GAs). Building on this foundation, a path-following neural network controller for the FWOMR is proposed, ensuring continuous and asymptotic stability of the system. Moreover, the proposed control framework demonstrates robust resilience against various disturbances and uncertainties, including those arising from variable environmental conditions, dynamic changes in system parameters, and fluctuations in wheel friction. The results of this study indicate that the control and navigation system maintain a high level of stability across diverse operational scenarios, as validated through both comparative simulations and experimental evaluations. In conclusion, the consistent performance observed under multiple conditions highlights the efficacy of the proposed approach, confirming its reliability for practical application.</p>