The paper "Learning-based Methods for Adaptive Informative Path Planning" by Marija Popović, Joshua Ott, Julius Rückin, and Mykel J. Kochenderfer explores the integration of learning-based methods into adaptive informative path planning (AIPP) to enhance adaptability, versatility, and robustness in robotics. The authors provide a comprehensive survey of current research, establishing a unified mathematical problem definition for AIPP and presenting two complementary taxonomies: one based on learning algorithms and another on robotic applications. They discuss the synergies, recent trends, and benefits of learning-based methods in AIPP frameworks, while also highlighting key challenges and future directions. The paper includes a catalog of reviewed papers, including publicly available repositories, to facilitate future studies in the field. The authors emphasize the importance of learning-based methods in addressing the complexities of AIPP, such as modeling and predicting new information in unknown environments, balancing exploration and exploitation, and handling uncertainties in sensor measurements and actuation. They also review different mapping methods, evaluation metrics, and benchmarks used in AIPP, and provide an overview of supervised learning, reinforcement learning, imitation learning, and active learning techniques. The paper aims to provide a comprehensive understanding of the state-of-the-art in learning-based methods for AIPP and their applications, as well as to identify potential avenues for future research.The paper "Learning-based Methods for Adaptive Informative Path Planning" by Marija Popović, Joshua Ott, Julius Rückin, and Mykel J. Kochenderfer explores the integration of learning-based methods into adaptive informative path planning (AIPP) to enhance adaptability, versatility, and robustness in robotics. The authors provide a comprehensive survey of current research, establishing a unified mathematical problem definition for AIPP and presenting two complementary taxonomies: one based on learning algorithms and another on robotic applications. They discuss the synergies, recent trends, and benefits of learning-based methods in AIPP frameworks, while also highlighting key challenges and future directions. The paper includes a catalog of reviewed papers, including publicly available repositories, to facilitate future studies in the field. The authors emphasize the importance of learning-based methods in addressing the complexities of AIPP, such as modeling and predicting new information in unknown environments, balancing exploration and exploitation, and handling uncertainties in sensor measurements and actuation. They also review different mapping methods, evaluation metrics, and benchmarks used in AIPP, and provide an overview of supervised learning, reinforcement learning, imitation learning, and active learning techniques. The paper aims to provide a comprehensive understanding of the state-of-the-art in learning-based methods for AIPP and their applications, as well as to identify potential avenues for future research.