A Survey of Imitation Learning Methods, Environments and Metrics Nathan Gavenski, Felipe Meneguzzi, Michael Luck, and Odinaldo Rodrigues present a comprehensive survey of imitation learning, a machine learning approach where an agent learns by imitating a teacher's behavior. This method offers a balance between trial-and-error learning and supervised learning, allowing agents to acquire complex behaviors without explicitly defining reward functions. The survey addresses challenges in comparing and evaluating imitation learning methods, environments, and metrics, proposing novel taxonomies for these areas. It discusses key issues, open challenges, and future directions in the field, emphasizing the need for standardized evaluation processes and environments. The survey reviews recent advancements in imitation learning, including model-based and model-free methods, and introduces new taxonomies for methods, environments, and metrics. It highlights the growing applications of imitation learning in robotics, game-playing, and natural language processing. The survey also discusses the role of demonstrations, experiences, and observations in imitation learning, and the differences between stationary and non-stationary policies. The survey categorizes imitation learning methods into five main types: behavioural cloning, inverse dynamics models, forward dynamics models, adversarial learning methods, and hybrid approaches. It discusses the advantages and limitations of each method, and how they can be combined to improve performance. The survey also addresses the challenges of generalization, sample efficiency, and the need for robust policies in dynamic environments. The survey concludes that imitation learning is a rapidly evolving field with significant potential for future research and applications. It emphasizes the importance of standardized evaluation processes and environments, and the need for further research into the challenges of imitation learning, such as resilience and evaluation processes. The survey provides a solid foundation for understanding the diverse and dynamic nature of imitation learning and offers a systematic approach for comparing and evaluating different techniques.A Survey of Imitation Learning Methods, Environments and Metrics Nathan Gavenski, Felipe Meneguzzi, Michael Luck, and Odinaldo Rodrigues present a comprehensive survey of imitation learning, a machine learning approach where an agent learns by imitating a teacher's behavior. This method offers a balance between trial-and-error learning and supervised learning, allowing agents to acquire complex behaviors without explicitly defining reward functions. The survey addresses challenges in comparing and evaluating imitation learning methods, environments, and metrics, proposing novel taxonomies for these areas. It discusses key issues, open challenges, and future directions in the field, emphasizing the need for standardized evaluation processes and environments. The survey reviews recent advancements in imitation learning, including model-based and model-free methods, and introduces new taxonomies for methods, environments, and metrics. It highlights the growing applications of imitation learning in robotics, game-playing, and natural language processing. The survey also discusses the role of demonstrations, experiences, and observations in imitation learning, and the differences between stationary and non-stationary policies. The survey categorizes imitation learning methods into five main types: behavioural cloning, inverse dynamics models, forward dynamics models, adversarial learning methods, and hybrid approaches. It discusses the advantages and limitations of each method, and how they can be combined to improve performance. The survey also addresses the challenges of generalization, sample efficiency, and the need for robust policies in dynamic environments. The survey concludes that imitation learning is a rapidly evolving field with significant potential for future research and applications. It emphasizes the importance of standardized evaluation processes and environments, and the need for further research into the challenges of imitation learning, such as resilience and evaluation processes. The survey provides a solid foundation for understanding the diverse and dynamic nature of imitation learning and offers a systematic approach for comparing and evaluating different techniques.