Dilli Bhaskar

M.A.Sc Candidate, Electrical & Computer Engineering, University of British Columbia

I'm a 1st year master's student at UBC, advised by Prof. Cyrus Neary. Previously, I was a Research Intern at IIT Palakkad working on Deep RL-based motion planning with Prof. Shaikshavali Chitraganti. I completed my Bachelor's in Mechanical Engineering at TCE under the guidance of Prof. C. Selva Kumar.

I'm interested in leveraging reinforcement learning and optimal control to build robots that learn and adapt.

How can deep RL enable robust autonomous navigation and maneuver execution?
Can model-free entropy frameworks improve sample efficiency and generalization in continuous control?
How do we bridge sim-to-real transfer for multi-agent systems?

Email | CV | GitHub | LinkedIn | Google Scholar

Highlights

 New Two papers accepted to the 11th Indian Control Conference (ICC), Bangalore, 2025. 
Starting M.A.Sc at University of British Columbia (Sep 2025), advised by Prof. Cyrus Neary. 
Paper on congestion-aware path planning published in Industry 4.0 and Advanced Manufacturing, Springer. 
Joined IIT Madras as a Project Associate working on pursuit-evasion UAVs with Deep RL. 
Graduated B.E. Mechanical Engineering from TCE, Madurai — Department Best Outgoing Student. 
U.K. Design Patent filed: "Robotic Duct Vacuum Cleaner" (No. 6339408). 
Ranked Top 6 in the DePondFi '23 Challenge at NCVPRIPG-2023. 
Paper on thermal fire detection accepted at IC3I 2022. 

Research

Soft Normalized Advantage Functions for Reinforcement Learning in Optimal Control Problems

Dilli Bhaskar, Shaikshavali Chitraganti

Accepted to 11th Indian Control Conference (ICC), Bangalore, 2025

Value-Based Reinforcement Learning for Mapless Navigation of a Mobile Robot

Dilli Bhaskar, Ayyappadas Rajagopal, Nandagopan Kalathil, Shaikshavali Chitraganti

Accepted to 11th Indian Control Conference (ICC), Bangalore, 2025

Video

Congestion-aware Path Planning for Multiple Shelf Carrying Mobile Robots in RMFS

Dilli Bhaskar, Selva Kumar Chandrasekar, Subramanian Saravana Perumaal

Industry 4.0 and Advanced Manufacturing, Volume 1, Springer, 2024

Paper

Early Thermal Forest Fire Detection using UAV and Saliency Map

Dilli Bhaskar, Suguna M.

5th International Conference on Contemporary Computing and Informatics (IC3I), 2022

Paper

Performance Enhancement of RMFS with Heterogeneous Robots and Soft Q-Learning-Based Order Allocation

Dilli Bhaskar, Selva Kumar Chandrasekar, Subramanian Saravana Perumaal

Under Journal Review

Projects and Competitions

Below are some of the projects I had fun working on, click the title for details.

Robotic Manipulation using Continuous Deep RL

Control and Robotics Lab, IIT Palakkad

Developed DRL pipeline to train a 7-DOF robotic arm in PandaReach-v3 using panda-gym. Implemented SAC, NAF, and TD3 with Hindsight Experience Replay to address sparse rewards.

GitHub

Aircraft Prototype Design

Team Yukta Racing, TCE

Designed and fabricated aircraft prototypes with a focus on aerodynamic efficiency and structural integrity. Conducted flight testing and iterative design improvements using CAD tools and rapid prototyping techniques.

PondFishDet

DePondFi'23, NCVPRIPG-2023

Developed CLAHE-YOLOv8 and MSR-YOLOv8 algorithm to enhance underwater image analysis for accurate fish detection. Achieved MAP scores of 0.964 and 0.970. Ranked Top 6 in the DePondFi '23 Challenge.

GitHub

Street Litter Detection

Hack4Good Hackathon 2022

Developed a street litter detector using YOLOv5 based on Euclidean approach. Trained on Pennfudan and TACO Dataset. Achieved F1 score of 89%. Successfully deployed on real-time CCTVs.

GitHub Video

LQR Controlled Self Balancing Bike

E-Yantra Robotics Competition | Placed 18/198

Developed and implemented a specialized LQR controller for a single reaction wheel-based self-balancing robot. Verified through simulations in V-rep. Constructed the bike using Solidworks 2023.

GitHub Video