Autonomous navigation of a low-cost service robot in a dynamic restaurant environment using enhanced SLAM and GA-optimized fractional PI control

• Unified framework of odometry, SLAM, path planning, and GA-FOPI control. • Robust RGB-D odometry using SURF, NNDR, PnP, and RANSAC. • EKF-based SLAM fusing IMU and RGB-D to reduce drift and noise. • Hybrid A*–DWA for optimal global and local navigation. • Validated in simulation and real restaurant environment. This paper addresses critical challenges in autonomous service robotics for a low-cost restaurant robot equipped with a 3D camera. The proposed framework integrates: (i) a vision-based odometry method using RGB-D data for accurate pose estimation, (ii) an enhanced SLAM system fusing IMU and RGB-D measurements via the Extended Kalman Filter to reduce drift and noise, (iii) a hybrid path planning scheme combining A* for global route optimization with the Dynamic Window Approach for real-time local navigation, and (iv) a Genetic Algorithm–optimized fractional-order PI (GA-FOPI) controller for precise trajectory tracking. Experimental and simulation results show that the GA-FOPI outperforms conventional PI and FOPI controllers in speed, accuracy, and stability. Specifically, the rise time was reduced from 0.35 s to 0.17 s, the settling time from 0.58 s to 0.26 s, and the overshoot nearly eliminated (0.01 %). The integrated system achieves high mapping accuracy, robust navigation, and reliable control in a dynamic restaurant environment. The novelty lies in unifying these techniques into a single cost-effective ROS-based platform, validated through both simulation and real-world deployment, a combination not previously reported in the literature.