Implementasi Algoritma Cuckoo Search untuk Optimasi Lintasan dan Penghindaran Rintangan Robot PoleBot

Abstract

This study implements the Cuckoo Search Algorithm (CSA) for path optimization and obstacle avoidance on the differential-drive PoleBot AMR running ROS2 Jazzy. CSA is applied at two levels: a global CSA that optimizes waypoints for three candidate paths, and a local CSA that selects linear and angular velocity pairs over a 1.15 s prediction horizon as the obstacle-avoidance layer. Both run as a single ROS2 node integrated with the Tongyi servo motor driver (CANopen), the Autonics LSC 270° 2D LiDAR, and AMCL localization. The global CSA planner was compared with the Genetic Algorithm (GA) and RE-RRT on an identical navigator with the CSA local layer fixed, in 30 2D-simulation runs (10 per configuration). All configurations reached the goal without collision. GA was the most efficient (17.4 m; 177 s) with the smallest minimum clearance (0.59 m); RE-RRT produced the longest and most variable paths (19.4 ± 1.8 m) with the largest minimum clearance (1.69 m); CSA lay in between (17.6 m; 181 s; 0.67 m). On the physical robot with static obstacles, 10 of 10 runs succeeded without collision (242 ± 32 s; path efficiency 0.84). The physical implementation exposed six integration problems, including stale TF poses and the motor braking limit, together with their fixes.

Keywords— Autonomous Mobile Robot, Cuckoo Search Algorithm, Obstacle Avoidance, PoleBot, ROS2