About Me

I’m a graduate researcher at the KTH Royal Institute of Technology, finishing my MSc in Systems, Control & Robotics with a thesis at ABB Robotics on generative models for long-horizon manipulation under formal specifications.

My current work uses flow matching steered by Signal Temporal Logic robustness gradients to plan robot trajectories that satisfy task specifications without retraining for each new task. Before this, I worked on 3D scene graphs at Ericsson Research (lead author on two submissions to IMPROVE 2026 and ICPR 2026), and on deformable object manipulation with vision-language models at the KTH Robotics, Perception & Learning lab (presented at ROMADO @ IROS 2025).

I’m drawn to methods that compose - small, well-understood building blocks combined at inference time, rather than monolithic policies retrained per task. Before Sweden, I completed a Dual Degree (B.Tech Engineering Design + M.Tech Robotics) at IIT Madras, with hands-on experience deploying real robotic systems at Systemantics, Yaskawa, and Imaginate.

Research Interests

News

Publications

For the full publication list, see my Google Scholar.

Scene Understanding in Deformable Object Manipulation via Taxonomy-Guided Vision-Language Models Gawtam Chithra Ramesh, David Blanco-Mulero, Yifei Dong, Júlia Borràs Sol, Carme Torras, Florian T. Pokorny IROS 2025 ROMADO Workshop arXiv project page code

Robust Object-layer Construction of 3D Scene Graphs Using Instance Segmentation Gawtam Chithra Ramesh, Püren Güler, Hiba Alqaysi, Marcus Valtonen Örnhag1, Héctor Caltenco, and Erdal Akin IMPROVE 2026 (Accepted) arXiv paper – to appear

Synchronized RGB-D Inpainting for Privacy-Aware 3D Scene Graph Construction Gawtam Chithra Ramesh, Püren Güler, Hiba Alqaysi, Marcus Valtonen Örnhag1, Héctor Caltenco, Kayode Sakariyah Adewole, and Erdal Akin ICPR 2026 (Under Review) arXiv

Projects

STL-Guided Flow Matching for Long-Horizon Manipulation

Flow Matching Signal Temporal Logic Manipulation PyTorch MuJoCo

code – coming soon

Open-Vocabulary 3D Semantic Mapping + VLA Deployment

OwlV2 SAM CLIP VLA RGB-D PyTorch

Code

Vision-Language Models for Deformable Object Manipulation

Vision-Language Models Gemini Qwen Cloth Manipulation Taxonomy

project page Code

Multi-Agent Path Finding in Unity3D

MAPF Conflict-Based Search Hybrid A* Unity3D C#

Code

RL + Behavior Trees for Pacman Capture-The-Flag

Reinforcement Learning PPO Behavior Trees Unity ML-Agents

Code

6-DoF Manipulator Trajectory Generation @ Systemantics

C++ ROS2 SocketCAN D-Bus Behavior Trees Real-Time Control

Code – coming soon

Work Experiences

Skills

Honors and Awards

Services and Responsibilities

Educations

08/2024 - 07/2026

Master of Science in Systems, Controls, and Robotics

KTH Royal Institute of Technology, Sweden

08/2019 - 07/2024

Dual Degree (B.Tech in Engineering Design and M.Tech in Robotics)

Indian Institute of Technology Madras, India