<![CDATA[The Indonesian Wheeled Soccer Robot Contest requires autonomous robots to navigate safely in dynamic environments where opponent robots continuously change their positions. Conventional Single A-star path planning generates the navigation path only once before robot movement, making it unable to adapt to obstacle movement during navigation. This study proposes a ROS 2-based Dynamic A-star path planning system for opponent avoidance in wheeled soccer robots. The proposed system performs real-time path replanning whenever the detected obstacle position changes, enabling the robot to generate an alternative collision-free trajectory. The system was evaluated in the Gazebo simulation environment using eight navigation scenarios and compared with the conventional Single A-star approach. The same robot starting positions and goal locations were used in both approaches, while the Dynamic A-star method extended the baseline configuration from one obstacle to two obstacles, including one moving obstacle. Experimental results showed that the proposed Dynamic A-star method achieved a 100% navigation success rate with a 0% collision rate, whereas the conventional Single A-star algorithm achieved a 0% navigation success rate with a 100% collision rate after obstacle movement. The average travel time increased from 19.50 s for Single A-star to 29.95 s for Dynamic A-star, while the average path length increased from 4.08 m to 6.76 m due to additional detours and path replanning. These results demonstrate that Dynamic A-star improves navigation adaptability and collision avoidance in dynamic environments. In addition, the ROS 2 framework provided real-time communication among perception, path planning, and robot control modules, supporting adaptive autonomous navigation for wheeled soccer robots.]]>
