Wenshan Wang

Carnegie Mellon University

Papers

4

Total Citations

105

H-Index

3

About

Wenshan Wang is a robotics researcher specializing in off-road autonomous navigation, self-supervised learning, and robot perception. His work centers on enabling robots to intelligently interpret and traverse complex, unstructured terrains — a notoriously difficult challenge in field robotics. Wang's most influential contribution, "How Does It Feel?" (2023, 56 citations), introduced a self-supervised costmap learning framework that eliminates the need for hand-crafted terrain labels by leveraging robot-terrain interaction dynamics directly. Complementing this, his work on risk-aware costmaps via inverse reinforcement learning (2023, 23 citations) advances how robots learn safe navigation strategies from expert demonstrations, reducing costly engineering overhead in costmap design. A recurring theme in Wang's research is building robust infrastructure for the broader community. His TartanDrive dataset series — with version 2.0 (2024, 25 citations) expanding modalities and scale — has become a foundational resource for off-road self-supervised learning research. His most recent work, Tartan IMU (2025), pushes toward generalizable foundation models for inertial odometry, addressing critical limitations in real-world deployment. Collectively, Wang's research shapes how autonomous systems perceive, learn from, and safely navigate challenging real-world environments, making him a significant contributor to field and off-road robotics.

Research Focus

Key Achievements

3
H-Index
4
Papers
105
Total Citations
26
Avg Citations/Paper
🏆 Most Cited Paper
How Does It Feel? Self-Supervised Costmap Learning for Off-Road Vehicle Traversability
56 citations · 2023
📈 Most Prolific Year: 2023 (2 Papers)
🤝 Key Collaborators: 14
🏛 Institutions: Carnegie Mellon University

Top Papers

  1. 1
  2. 2
  3. 3
  4. 4

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 14 days ago