Generalization
Related papers: 20
About
Generalization in robotics and AI refers to a system's ability to apply knowledge, skills, or policies learned from training data or experiences to new, previously unseen situations, environments, or tasks. Rather than simply memorizing specific examples, a generalizing system extracts underlying patterns or principles that transfer broadly. In practice, generalization appears across many robotics and AI domains: a robot trained to grasp familiar objects must handle novel shapes and textures; a navigation agent learned in one environment must perform in another; a motor skill demonstrated by a human must be adapted to varied physical conditions. Techniques such as deep reinforcement learning, learning from demonstration, dynamic movement primitives, and statistical learning theory all directly address how well-learned representations scale beyond their original training distribution. Generalization matters profoundly because real-world deployment is inherently open-ended. Robots encounter unpredictable variability in objects, environments, and task requirements. Systems that fail to generalize require expensive retraining for every new scenario, severely limiting practical utility. Achieving robust generalization is therefore considered one of the central unsolved challenges in robotics and AI, directly determining whether learned capabilities remain reliable and useful outside controlled laboratory conditions.
Top Researchers
Top Institutes
Top Cited Papers
Statistical Learning Theory
Yuhai Wu, Vladimir Vapnik
Citations: 26957 • 1999
Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators
Citations: 2732 • 2021
Trends in extreme learning machines: A review
Gao Huang, Guang-Bin Huang, Shiji Song, Keyou You
Citations: 1738 • 2014
Target-driven visual navigation in indoor scenes using deep reinforcement learning
Yuke Zhu, Roozbeh Mottaghi, Eric Kolve, Joseph J. Lim, Abhinav Gupta, Li Fei-Fei, Ali Farhadi
Citations: 1507 • 2017
An Introduction to Deep Reinforcement Learning
Vincent François-Lavet, Peter Henderson, Riashat Islam, Marc G. Bellemare, Joëlle Pineau
Citations: 1246 • 2018
Supersizing self-supervision: Learning to grasp from 50K tries and 700 robot hours
Lerrel Pinto, Abhinav Gupta
Citations: 1099 • 2016
Reinforcement learning for robots using neural networks
Long-Ji Lin
Citations: 887 • 1992
Scan registration for autonomous mining vehicles using 3D‐NDT
Martin Magnusson, Achim J. Lilienthal, Tom Duckett
Citations: 767 • 2007
Learning and generalization of motor skills by learning from demonstration
Peter Pástor, H. Hoffmann, Tamim Asfour, Stefan Schaal
Citations: 710 • 2009
Control Theory of Non-linear Mechanical Systems
Suguru Arimoto
Citations: 580 • 1996
Cooperative Robot Control and Concurrent Synchronization of Lagrangian Systems
Soon‐Jo Chung, Jean-Jacques Slotine
Citations: 551 • 2009
An Introduction to Deep Reinforcement Learning
Vincent François-Lavet, Peter Henderson, Riashat Islam, Marc G. Bellemare, Joëlle Pineau
Citations: 539 • 2018
Rapid Biologically-Inspired Scene Classification Using Features Shared with Visual Attention
Christian Siagian, Laurent Itti
Citations: 525 • 2007
Lifelong robot learning
Sebastian Thrun, Tom M. Mitchell
Citations: 507 • 1995
Emergence of Functional Hierarchy in a Multiple Timescale Neural Network Model: A Humanoid Robot Experiment
Yuichi Yamashita, Jun Tani
Citations: 487 • 2008
Coverage of Known Spaces: The Boustrophedon Cellular Decomposition
Howie Choset
Citations: 482 • 2000
Understanding sensorimotor adaptation and learning for rehabilitation
Amy J. Bastian
Citations: 481 • 2008
Real-World Multiobject, Multigrasp Detection
Fu-Jen Chu, Ruinian Xu, Patricio A. Vela
Citations: 447 • 2018
Control-limited differential dynamic programming
Yuval Tassa, Nicolas Mansard, Emo Todorov
Citations: 441 • 2014
Robotic neurorehabilitation: a computational motor learning perspective
Vincent Huang, John W. Krakauer
Citations: 424 • 2009