Bilas Talukdar

Latest: 

March 2025

• Our paper titled "Generalized, Dynamic Multi-agent Torso Crowds" has been accepted at "2025 ACM SIGGRAPH Symposium on Interactive 3D Graphics and Games (I3D 2025)".

In this work, we propose novel position-based dynamics (PBD) constraints for capsule crowd navigation. Capsules better approximate the profile of the human body compared to traditional disc-shapes or ellipses.  We represent each crowd agent with a torso, which realistically captures a pedestrian’s rotation and shape. This enables simulating real-world dynamic details that are difficult to capture with discs. Our approach works in settings ranging from sparse to dense

May 2024

• I have presented our paper titled "Learning Crowd Motion Dynamics with Crowds" at "2024 ACM SIGGRAPH Symposium on Interactive 3D Graphics and Games (I3D 2024)", Philadelphia, PA, USA.

February 2024

• Our paper titled "Learning Crowd Motion Dynamics with Crowds" has been accepted at "2024 ACM SIGGRAPH Symposium on Interactive 3D Graphics and Games (I3D 2024)".

In this work, we proposed a crowd dynamics framework combining RL and position-based dynamics which propel agents to move in human-inspired fashion.  In addition, we propose a crowd-sourced, Bayesian approach for selecting optimal learning parameters.

August 2023

• One of our papers got accepted at the "Smoky Mountains Computational Sciences and Engineering Conference (SMC2023)" in the Data Challenge session.

In this work, we investigated five different questions regarding power outage data across the United States at the county level. This data was collected through the Environment for the Analysis of Geo-Located Energy Information (EAGLE-I) platform at Oak Ridge National Laboratory (ORNL). Thanks to my collaborators for their contributions.

June 2023

• I’m excited to share that I’m starting a new position as a Summer Research Intern 2023 at the Oak Ridge National Laboratory. I'll work on updating regression test coverage for the world's first and fastest exascale supercomputer, Frontier, to further extend coverage of AI and ML benchmarks.

• I am happy to share that our two-page abstract titled "Learning to Simulate Crowds with Crowds", has been accepted to the prestigious conference SIGGRAPH 2023 in the posters category. This will be presented as a poster in the SIGGRAPH 2023 in Los Angeles, California in August, 2023. Also, the abstract and poster will be published in the ACM Digital Library. This year, the acceptance rate of the abstracts/posters is 35.7%.

November 2022

• One of my recent works' ("A Deep Reinforcement Learning Approach Towards Human-Like Multi-Agent Dynamics") abstract has been accepted at 2022 SIAM Conference on Mathematics of Data Science; September 2022; San Diego, California, U.S. You can find the abstract here.

November 2022

• I worked as a Summer Research Intern at the Argonne National Laboratory (June 2022 - August 2022). See more about this.

• One of my recent works' ("CrowdRL: Training First Responders for Complex Crowd Interaction Scenarios with Reinforcement Learning") abstract has been accepted at NJBDA Symposium 2022.