Physics-trained. Robot-focused. I build intelligent agents that move, sense and adapt, from robot kinematics to learned control policies. MSc AI at Edinburgh, 2025-2026. SLMC group, IPAB. Earlier research at IIT Delhi and ISRO.
I'm a Research Assistant in the Statistical Machine Learning and Motor Control (SLMC) group at IPAB, University of Edinburgh, focused on robot learning, dexterous manipulation, tactile sensing and sim-to-real reinforcement learning. My background in AI and Engineering Physics brings together mathematics, signal processing and dynamical systems. I apply that foundation to teaching robots how to move, perceive and make decisions.
I'm drawn to embodied intelligence: agents that act in the physical world under uncertainty. From processing satellite data at ISRO to computational ocean analysis at IIT Delhi, I've always worked at the edge of computation and physical systems. Robotics is where those worlds meet permanently.
I've built web apps, competed in science hackathons (Top 20 nationwide in XR-tech), and led teams. I believe the best engineers are curious generalists who execute with precision.
Advanced Robotics; Robot and Reinforcement Learning; Probabilistic Modeling and Reasoning; Advanced Topics in Machine Learning; Computational Neuroscience; Computational Cognitive Neuroscience; Introduction to Quantum Computing; Computer Vision.
Fundamental Physics; Mathematics; MATLAB Applied Physics; Quantum Physics; Cosmology and Astrophysics; Electronics; Semiconductors; Optical Fibres; Microwave; Atmospheric Sciences; Data Structures and Algorithms; Database Systems; Operating Systems; Computer Networks; Computer Graphics.
Simulation, control and robot interfaces for manipulation. Current research in the Statistical Machine Learning and Motor Control group at the University of Edinburgh.
Real-time environments & object resets
Parallel sweeps with mjbatch
Trajectories, teleoperation & cameras
MuJoCo simulation infrastructure for real-time box folding, object-state resets and trajectory execution. Controller-gain tuning and parallel parameter sweeps with mjbatch connect simulation to reinforcement-learning and data-collection pipelines.
ROS2 interfaces for trajectory editing and folding execution on NEXTAGE. Streaming teleoperation and simultaneous publishing from multiple robot cameras support interaction and data collection.
MuJoCo model integration and controller stabilisation for the AIREC robot. A focus on the simulation and control foundations needed for reliable robot behaviour.
The tactile system combines 12 FSR sensors and four BioTac fingertips with per-channel hysteresis calibration. PadTac USD assets in Isaac Lab, a ROS deployment pipeline and slew-rate domain randomisation account for sensing and hardware motion limits. PPO policies transfer zero-shot to the physical hand.
MSc dissertation, Jan-Sept 2026: Development of a Low-cost Tactile System for Dexterous Manipulation and Investigating the Sim-To-Real Gap in Tactile Reinforcement Learning.
Supervision: Prof. Sethu Vijayakumar and Elle Miller.
A Novel Dielectric modulated Double Gate Charge-Trapping MOSFET: Ferroelectric Gate Stack and Gaussian Doped Substrate Integration for Enhanced Performance.
Akkshita Swain, Nalin Agarwal, Tanisha Singhal, Rishu Chaujar.
Undergraduate research: Aug 2023-May 2024, Delhi Technological University.
↗ Conference certificateOn Rapid Increase In The MHW Intensity During Last Decade (2010-Present) Using Satellite Observations: Implications on Cyclone Activity in the Arabian Sea?
Nalin Agarwal, Nitant Dube, Rashmi Sharma.
Presented at IISER Bhopal, December 2022.
↗ Presentation certificateLeading design teams and shaping print and digital work, from brochures and event identities to websites and social media graphics.
Fundraising, awareness drives and conversations about the foundation's work for children in India.
↗ Volunteer certificateContent research and corrections, blog posts, website landing pages and digital marketing with SEO. Turning research into clear communication and accessible web content.
↗ Content research certificateSchool topper in Class X board examinations · Academic Head Boy in Class XII
My long-term goal is to build embodied systems that could develop subjective experience and a sense of self. I want to understand how consciousness and self-awareness could emerge through perception, memory and action.
I want to explore agents that build a model of themselves, reflect on what they know and learn through experience in the physical world. Robotics gives me a way to connect those questions to perception, action and learning.
Explore how an embodied agent could learn the relationship between its body, its actions and the world around it.
Investigate how memory and a continuing history of experience could shape an agent's model of itself over time.
Study how agents might track uncertainty, recognise their own limits and use that information to guide learning.
If you're working on robotics, robot learning or RL, I'd love to hear from you. I bring a physics and systems thinking background to research and collaboration.