Gyan Tatiya
Papers
8
Total Citations
79
H-Index
5
About
Gyan Tatiya is a robotics and artificial intelligence researcher whose work sits at the compelling intersection of multi-sensory perception, knowledge transfer, and robot learning. Drawing inspiration from human cognition, Tatiya investigates how robots can leverage multiple sensory modalities — vision, audio, and haptics — alongside exploratory behaviors such as grasping, pushing, and shaking to develop richer, more complete understandings of object properties. His most influential contribution, "Deep Multi-Sensory Object Category Recognition Using Interactive Behavioral Exploration" (2019, 26 citations), introduced a deep learning framework enabling robots to recognize object categories through integrated visual, auditory, and haptic data, mirroring how humans naturally discover the physical world. Subsequent work tackled the challenge of transferring this hard-won sensorimotor knowledge across robots with different morphologies and sensory configurations, a significant practical barrier in real-world robotics deployment, accumulating over 35 additional citations across multiple publications. Tatiya has also explored curriculum-based reinforcement learning through ACuTE, addressing the steep sample complexity that hampers real-world task learning. His most recent work, MOSAIC, advances unified multi-sensory object property representations, pushing toward robots capable of genuinely holistic environmental understanding. Collectively, his research charts an ambitious path toward robots that perceive and learn about their world as intuitively as humans do.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5
- 6ACuTE: Automatic Curriculum Transfer from Simple to Complex Environments4 citations · 2022
- 7
- 8