Cruz, FranciscoSolis, Miguel A.Navarro-Guerrero, Nicolás2024-05-232024-05-232023Frontiers in Neurorobotics. Volume 17. 2023. Article number 12567881662-5218https://repositorio.unab.cl/handle/ria/57033Indexación: Scopus.Robot learning enables robots to acquire new knowledge and skills through experience and interaction with their environment. Robot learning involves developing algorithms that allow robots to learn autonomously, adapt to new situations, and improve their performance over time. Using machine learning, robots can analyze large amounts of data and extract patterns to make decisions. This approach allows robots to learn from past experiences and apply that knowledge to future tasks, ultimately enhancing their capabilities and versatility. However, although machine learning has shown great potential in robot learning, it also faces several challenges and limitations. One significant problem, for instance, is the issue of data scarcity. Collecting sufficient and diverse data for training robots can be complex and time-consuming (Navarro-Guerrero et al., 2023). Unlike traditional machine learning applications where large datasets might be available, gathering data for robot learning often requires physical interactions and real-world environments, which can be expensive and challenging (Navarro-Guerrero et al., 2023).enBio-inspired roboticsCognitive roboticsRobot learning and behavior adaptationRoboticsSocial robotics/HRI engineersArtículoCC BY 4.0 DEED Attribution 4.0 International10.3389/fnbot.2023.1256788