Planejamento de rotas em ambientes dinâmicos: uma comparação entre algoritmos na RoboCup SSL

Abstract

This work presents a comparative analysis between classical and incremental path planning algorithms applied to the RoboCup Small Size League (SSL), an environment characterized by high dynamics, multiple robots, and frequent physical interactions among them. In this context, the choice of the path planner directly influences navigation safety, trajectory efficiency, and the overall team performance, which justifies the need for a structured evaluation of different approaches. The algorithms Dijkstra, A*, Lifelong Planning A (LPA*), and D* Lite were implemented and compared, all operating on the same Visibility Graph representation to ensure a common basis for analysis. The experiments reproduce typical game situations involving both static and dynamic obstacles, allowing the observation of how each method responds to varying levels of environmental demands. The evaluation combined qualitative visual analyses and quantitative performance metrics, along with full-match simulations, highlighting clear differences between classical and incremental planners, particularly regarding adaptation to environmental changes and collision occurrence. The results provide a consistent understanding of the advantages and limitations of each approach and offer practical support for selecting path planning strategies in mobile robotic systems operating under dynamic conditions.