Paper Contents
Abstract
Transfer learning has emerged as a transformative approach in robotics by which the robot uses past experiences to enhance the efficiency and adaptability of the performance of new tasksenvironments. In this paper, we shall consider how TL is benefitting its applications in robotics: perception, manipulation, navigation, and human-robot interaction. TL allows robots to generalize across domains without having to restart data collection and severely long retraining on a different task, thereby providing the versatility required. Nonetheless, certain limitations curb its potential for full use, including domain mismatch, heavy computational requirements, and soft-level forgetting. A future research agenda for improved simulation-to-real gap transfer, cross-domain learning, and lifelong learning illustrates the centrality of TL in the future of adaptive and intelligent robotics.
Copyright
Copyright © 2025 Amrit chanana. This is an open access article distributed under the Creative Commons Attribution License.
