Atiksh Bhardwaj

About Me

Atiksh Bhardwaj

Hi, I'm Atiksh and am currently a Junior at Cornell University studying Computer Science. At Cornell, I'm part of the PoRTaL research group advised by Prof. Sanjiban Choudhury.

I enjoy finding new methods of teaching robots and working robotics systems and am interested in Imitation Learning, Reinforcement Learning, and Human-Robot Interaction.

Experience

(Summer 2023 - Present) Research Intern @ PoRTaL

(Fall 2024) TA for Robot Learning (CS 4756)

(Spring 2024) TA for Artificial Intelligence (CS 4700)

(Fall 2023) Consultant for Functional Programming (CS 3110)

(Summer 2022) Robotics Intern @ Brains4Drones

News and Media

(Oct. 16th 2022) NFTTRee wins PI Network Track at Big Red Hackathon

(Mar. 4th 2022) MathWorks Blog: Sibling Duo Share How Participating in Student Competitions Drives Interest in STEM Careers

Publications

MOSAIC Project Thumbnail

MOSAIC: A Modular System for Assistive and Interactive Cooking

Huaxiaoyue Wang*, Kushal Kedia*, Juntao Ren*, Rahma Abdullah, Atiksh Bhardwaj, Angela Chao, Kelly Y Chen, Nathaniel Chin, Prithwish Dan, Xinyi Fan, Gonzalo Gonzalez-Pumariega, Aditya Kompella, Maximus Adrian Pace, Yash Sharma, Xiangwan Sun, Neha Sunkara, and Sanjiban Choudhury

8th Annual Conference on Robot Learning (CoRL), 2024

Awards: Best Paper Award at the ICRA 2024 VLNMN Workshop and Best Poster Award at the ICRA 2024 MoMa Workshop

Project Page / Paper / Code

InteRACT Project Thumbnail

InteRACT: Transformer Models for Human Intent Prediction Conditioned on Robot Actions

Kushal Kedia, Atiksh Bhardwaj, Prithwish Dan, Sanjiban Choudhury, ICRA, 2024

IEEE International Conference on Robotics and Automation (ICRA), 2024

Project Page / Paper / Code

ManiCast Project Thumbnail

ManiCast: Collaborative Manipulation with Cost-Aware Human Forecasting

Kushal Kedia, Prithwish Dan, Atiksh Bhardwaj, Sanjiban Choudhury, CoRL, 2023

7th Annual Conference on Robot Learning (CoRL), 2023

Project Page / Paper / Code

Projects

DINK Project Thumbnail

DINK: Differently Initialized Q-Networks

Atiksh Bhardwaj, Jonathen Chen

Q-Networks represent an off-policy reinforcement learning approach centered on estimating value functions from environments. As an off-policy method, it relies on an external dataset for learning and computing Q-values. This dependency on provided data can yield diverse outcomes, either enhancing or hindering the Q-Network's performance. To delve into this dynamic, we embarked on an investigation utilizing conventional imitation learning techniques like Behavior Cloning and DAgger to generate data for training Q-Networks.

Video / Paper / Code