Yiwei Fu

Pennsylvania State University, GE Global Research (United States)

Papers

Total Citations

H-Index

About

Yiwei Fu is a robotics researcher whose work lies at the intersection of imitation learning, deep neural architectures, and autonomous systems. Their most influential contribution, the 2019 paper on neural network-based learning from demonstration for autonomous ground robots, has garnered 21 citations and introduces a novel end-to-end imitation learning framework. By combining convolutional neural networks (ConvNets) with Long Short-Term Memory (LSTM) networks, Fu created a spatio-temporal deep neural architecture that enables robots to learn complex behaviors directly from human demonstrations—eliminating the need for handcrafted features or explicit programming. This work was experimentally validated on real autonomous ground robots, demonstrating practical applicability in unstructured environments. Building on this foundation, Fu further advanced the field with their 2020 work on spatiotemporal representation learning using GAN-trained LSTM-LSTM networks. This unsupervised learning approach, called Layered Spatiotemporal Memory LSTM-LSTM, learns underlying robot behaviors without labeled data, pushing the boundaries of how autonomous systems can adapt to novel, unstructured settings. Fu’s research represents a significant step toward more flexible, data-efficient robot learning.