Gershom Seneviratne
Papers
3
Total Citations
8
H-Index
2
About
Gershom Seneviratne is a robotics researcher pushing the boundaries of autonomous navigation by bridging the gap between high-level human language and low-level physical control. His work centers on integrating Vision-Language Models (VLMs) and Large Language Models (LLMs) with physical grounding to enable robots to understand and act in complex, unstructured outdoor environments. Seneviratne’s major contributions include the development of novel algorithms like **BehAV**, which uses LLMs to parse human instructions into behavioral and navigational rules for outdoor robot guidance, and **VLM-GroNav**, a system that physically grounds VLM outputs to assess intrinsic terrain properties for safer traversal. He has also introduced **CROSS-GAiT**, a cross-attention-based framework that fuses visual and time-series sensor data to dynamically adapt quadruped robot gaits across challenging terrains. With his most cited work accumulating early citations, Seneviratne is establishing a strong footprint in embodied AI, demonstrating how multimodal perception and language understanding can be directly translated into robust, real-world robotic autonomy.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3