Wilbert Pumacay
Papers
4
Total Citations
33
H-Index
3
About
Wilbert Pumacay is an emerging robotics researcher whose work sits at the exciting intersection of robot manipulation, vision-language models (VLMs), and embodied AI. His research addresses some of the most pressing challenges in modern robotics: enabling robots to generalize across diverse environments, automate complex real-world tasks, and recover intelligently from failure. Pumacay's most cited contribution, "THE COLOSSEUM" (2024, 22 citations), introduces a rigorous benchmark for evaluating generalization in robotic manipulation — a critical bottleneck in deploying robots beyond controlled laboratory settings. His work on "Manipulate-Anything" (5 citations) explores leveraging VLMs to automate robot demonstrations at scale, directly addressing gaps in data quality and diversity that limit learning-based approaches. With "AHA" (4 citations), he tackles the underexplored challenge of failure detection and reasoning in open-world manipulation, equipping robots with a form of self-awareness. His "RoboPoint" project (2 citations) advances spatial affordance prediction, enabling VLM-driven robots to plan precise, reliable action points. Though early in his career, Pumacay has already established a cohesive research identity — building the tools, benchmarks, and models needed to make robust, generalizable robotic manipulation a reality.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4