Akademik Veri Yönetim Sistemi
2025 IEEE RAS Summer School on Multi-Robot Systems, Praha, Çek Cumhuriyeti, 29 Temmuz - 05 Ağustos 2025, (Özet Bildiri)
With the increasing use of mobile robots, ensuring safe and efficient navigation in complex environments with both stationary and moving obstacles is crucial. This project aims to address this challenge by developing a navigation stack that incorporates a reinforcement learning-based local planner. Unlike previous studies, the project will leverage a large number of diverse parallel training environments arranged in an easy-to-difficult curriculum. Simulation tools that support parallel training will be utilized to enhance the robots’ generalization capabilities across varying difficulty levels while significantly reducing training time. Through parallel training and diverse environments, the project aims to achieve a highly generalized and efficient navigation stack within a short time.