Research Interests
I have been doing research on pedestrian intention prediction in urban places using vision-based deep learning approach using graph deep learning and graph attention techniques. In my research, I tackled
social scene modeling and situational awareness of pedestrian.
The target outcome was a framework that learns about pedestrian intention and forecasts their future trajectories. This remains an essential task to the self-driving vehicles and social robots. Anticipating plausible pedestrian trajectories requires an understanding of exchanged influence that the physical
structure exerts on pedestrians intents and they influence each other given the social norms
and etiquette.
Motivated by the safe navigation concerns and the limited hardware capacity, my thesis focused
on designing an autonomous "Self-Growing Context-Aware Spatial Graph Network" that models
the environment with minimal engineering effort or human intervention. The thesis established
new approach for growing online graphs that are scalable based on pedestrian motion history,
their sight attention angle and the surrounding scene features.
My research interests also includes: multimodal sensing, Robotics Motion, Smart City surveillance systems, AI mobile edge applications.
I'm currently serving as reviewer for IEEE IoT-J and Neurocomputing journals. Besides that, I maintain my medium blog, writing about recent advancements in AI and deep learning, explaining theoretical and applicable aspects.