Peitao Cheng

About

Peitao Cheng is a leading researcher in visual place recognition (VPR) and simultaneous localization and mapping (SLAM), with a focus on robust perception for autonomous systems. His work bridges the gap between traditional computer vision and modern deep learning architectures. Cheng’s most cited paper, “Hybrid CNN-Transformer Features for Visual Place Recognition” (2022, 66 citations), introduces a novel fusion of convolutional neural networks and transformer models to overcome the limitations of CNNs in modeling spatial structures, significantly improving recognition under severe appearance and viewpoint changes. He further advanced the field with “Robust Loop Closure Detection Integrating Visual–Spatial–Semantic Information via Topological Graphs and CNN Features” (2020, 29 citations), which integrates visual, spatial, and semantic cues to enhance SLAM loop closure detection—a critical component for long-term robot navigation. His recent work, “Transformer-based descriptors with fine-grained region supervisions” (2023, 11 citations), pushes the boundaries of VPR by leveraging transformer-based descriptors for finer localization. With over 100 total citations, Cheng’s contributions are shaping the next generation of robust, real-world autonomous navigation systems.

Research Focus

Key Achievements

Top Papers

1. 1
   Hybrid CNN-Transformer Features for Visual Place Recognition

   66 citations · 2022
2. 2
   Robust Loop Closure Detection Integrating Visual–Spatial–Semantic Information via Topological Graphs and CNN Features

   29 citations · 2020
3. 3
   Transformer-based descriptors with fine-grained region supervisions for visual place recognition

   11 citations · 2023

Key Collaborators

Contact & Links