Portfolio - Suriya B

About Me

My expertise spans multi-robot coordination, autonomous navigation, and visual imitation learning. I've worked on cutting-edge projects including a Harvard Medical School-funded autonomous wheelchair and multi-robot fleet management using Open RMF. Currently, I'm experimenting with Action Chunking Transformers (ACT) for robotic manipulation on the SO-100 arm.

I've gained international research experience at West Pomeranian University in Poland (3D magnetic imaging sensors) and IIT Madras Research Park (autonomous wheelchair localization). My work has been recognized with the Judge's Choice Award at eYIC National Finals 2023-24 at IIT Bombay.

My goal is to build embodied intelligence that bridges the gap between simulation and real-world deployment—robots that can truly sense, reason, and act autonomously.

Skills

Software

Python

C++

Docker

Git

ROS 2

PyTorch

Fusion 360

Simulation

Isaac Sim

Isaac Lab

Mujoco

Gazebo

Foxglove Studio

Hardware

Jetson Boards

Stereo/Depth Cameras

IMU

3D LiDAR

Microcontrollers

Framework

ROS 2 (Humble/Jazzy)

MicroROS

Isaac ROS

Open-RMF

Move-it

Projects

Smart Navigation with Adaptive Intelligence

Developed and deployed a multi-robot coordination system using the Open RMF framework for collision-free operation in industrial environments. Integrated navigation stack with VDA 5050 standards for fleet communication, achieving ~30% improvement in effectiveness through real-life testing. Used Micro-ROS for low-level devices and Zenoh protocol for robust inter-robot communication.

Open RMF Micro-ROS VDA 5050 Zenoh Protocol Navigation Stack

Project Files Demo Video

Self-Driving Autonomous Wheelchair

Judge's Choice Award - eYIC 2024 (IIT Bombay)

Secured funding from Harvard Medical School to develop an autonomous wheelchair for individuals with upper-body weakness. Designed a low-cost, retrofit-table mechanism with vision-based perception using stereo depth cameras for real-time obstacle detection. Implemented dynamic floor mapping and automated lift access control for multi-level indoor navigation. Selected as top 10 finalist from 299 teams at eYIC National Finals 2023-24.

ROS 2 Stereo Depth Cameras Computer Vision SLAM Perception Pipeline

Project Files Demo Video

Visual Imitation Learning on SO-100 Arm

Ongoing Research (Sep 2025 - Present)

Experimenting with visual imitation learning on the SO-100 robotic arm to bridge the sim-to-real gap. Leveraging pre-training on smol VLA (Vision-Language-Action model) for efficient learning. Implemented and fine-tuned Action Chunking with Transformers (ACT) policy, achieving stable trajectory prediction for multi-stage stacking tasks. Conducted Domain Randomization experiments (varying lighting and camera poses) to analyze policy robustness against visual noise.

PyTorch Transformers (ACT) SO-100 Arm Vision-Language-Action Domain Randomization

Research Notes

Get In Touch

Email bsurriya.work@gmail.com

Location India

LinkedIn linkedin.com/in/b-surriya

Education B.Tech Robotics & Automation, Karunya University

Availability Open to Robotics & Autonomous Systems opportunities