Shubham Kamble – Portfolio

About Me

I am a Robotics Controls Engineer with over four years of academic research, teaching and professional experience in real‑time robotics software, optimisation and machine learning. My work focuses on transitioning advanced algorithms from simulation to physical deployment, building ML/RL pipelines and developing control frameworks for complex multi‑body robots. With strong mathematical foundations in system identification, optimal control and reinforcement learning, I strive to bridge the gap between theoretical research and real‑world robotics applications.

Experience

Sept 2024 – Present

Robotics Controls Engineer

Neura Robotics GmbH, Metzingen, Germany

Designed and implemented a dynamic identification framework for torque constants, rotor inertia, friction and inertial parameters. Developed Fourier‑series‑based trajectory generation under Riemannian manifold constraints. Created analytical inverse and forward kinematics solvers for humanoid robots. Deployed advanced friction compensation algorithms and optimised genetic algorithm routines. Spearheaded the developement of a generalized inverse kinematics solver.

Nov 2021 – Present

Student Research Assistant

Laboratory for Machine Tools and Production Engineering (WZL), RWTH Aachen

Built ML pipelines (RNEA+FCNN, LNN) for structured learning of 6‑DOF industrial robot dynamics. Integrated a Doosan Cobot M1013 with hardware peripherals on ROS for reinforcement‑learning‑based assembly automation. Developed a SIMULINK inverse dynamics model for friction identification. Designed a novel inertial parameter identification routine using low‑velocity excitation trajectories. Created explainable AI models using Transformers and Captum for torque prediction interpretability. Deployed research notebooks on RWTH JupyterHub with Docker.

Oct 2022 – Mar 2024

Student Assistant – Optimal Control & Technical Informatics Lab

Chair of Intelligent Control Systems, RWTH Aachen

Guided students in implementing control algorithms (PI, PID, MPC) on FloatShield apparatus using Arduino; Prepared lecture notes on the Pontryagin Maximum Principle. Graded and evaluated coursework for over 50 students. Supported laboratory sessions on Kalman Filters, Model Predictive Control and Receding Horizon Control.

Aug 2018 – Aug 2019

Product R&D Engineer (Mechatronic Systems)

Carnot Technologies, Mumbai, India

Conceptualised and prototyped automation systems. Developed functional prototypes aligned with ISO/IEC standards. Generated Bills of Materials to streamline sourcing.

Education

Oct 2020 – May 2024

M.Sc. in Robotic Systems Engineering

RWTH Aachen University, Germany

Optimal Control, Reinforcement Learning, Numerical Optimisation, Self‑driving Vehicles, Machine Learning, Computer Vision and Advanced robot Kinematics/Dynamics.

Aug 2014 – May 2018

B.Tech. in Mechanical Engineering

Veermata Jijabai Technological Institute (VJTI), Mumbai, India

Mechatronics, Automotive Engineering, Electrical Drives, Computational Fluid Dynamics and Finite Element Methods.

Skills

C++ & Python

High‑performance robotics, control algorithms and reinforcement learning development.

Control & Optimisation

Advanced robot Kinematics & Dynamics, Optimal Control, and Trajectory Optimisation techniques.

State Estimation & Bayesian Optimisation

Advanced state estimation (Kalman, Extended-Kalman, Unscented Kalman, and advanced Particle filtering), and Bayesian Optimisation techniques.

Machine Learning & Reinforcement LearningL

Building ML pipelines and applying reinforcement learning to robotics.

Frameworks & Tools

Linux, Docker, Git, Gitlab, ROS2, JAX, PyTorch, MATLAB, SIMULINK.

System Identification

Inertial parameters, rotor inertia, friction parameters, and elastic model estimation.

Soft Skills

Independent learner, critical thinker, problem solver and good communicator.

Projects

Motion Generation Library

Reusable C++ and Python library for analytical trajectory generation with linear, parabolic, cubic, quintic, trapezoidal, jerk-limited double-S, and waypoint spline profiles for robotics and CNC applications.

View on GitHub

Planning Algorithms Library

From-scratch C++ implementations of RRT, RRT*, RRT-Connect, PRM/PRM*, FMT*, BIT*, and RABIT* with a unified interface, obstacle handling, test coverage, documentation, and optional Python bindings.

View on GitHub

State Estimation Library

Compact, robotics-oriented C++ estimation framework implementing Bayes filter recursion, KF/EKF/UKF, particle filtering, and Gauss-Hermite Gaussian filtering with pluggable process and measurement models.

View on GitHub

Diffusion Models Platform

PyTorch DDPM research platform with configurable U-Net diffusion models, Gaussian diffusion utilities, AMP+EMA training pipeline, sampling scripts, and trajectory visualization for forward and reverse diffusion dynamics.

View on GitHub

RL Projects Platform

Hands-on reinforcement learning implementation suite covering dynamic programming, Monte Carlo, temporal-difference learning, DQN/DDQN, policy gradients, actor-critic methods, and model-based planning.

View on GitHub

Lagrangian Neural Network Research Platform

Lagrangian Neural Network Platform

Research platform for LNN/DeLaN/FeLaN dynamics modeling with reproducible training workflows in PyTorch and JAX, including double-pendulum baselines and Pinocchio-based dataset generation.

View on GitHub

Robot Viewer

Full-stack robot model visualization tool with a FastAPI backend and React/TypeScript frontend for parsing and exploring URDF, MJCF, and USD robot descriptions directly in the browser.

View on GitHub

Advanced Inverse Kinematics

Implemented an iterative redundancy-resolving 7-DOF inverse kinematics solver using pseudo-inverse Jacobians, null-space projection, and secondary task optimization for robust motion execution.

View on GitHub

Research Project Thesis: Condition Number Optimisation

Designed and optimized excitation trajectories to minimize regressor condition number, improving identifiability and numerical robustness in robot inertial parameter estimation.

View on GitHub

Seminar: Differential Dynamic Programming

Analyzed and presented the paper on online trajectory optimization for synthesizing and stabilizing complex behaviors using Differential Dynamic Programming.

View on GitHub

Seminar: Review on the AlphaGo RL algorithm

Studied and presented DeepMind's AlphaGo pipeline, focusing on policy/value networks, Monte Carlo Tree Search integration, and reinforcement learning strategy improvement.

View on GitHub

Seminar: Ethics in AI and Robotics

Evaluated ethical implications of AI-driven robotics systems, including responsibility, safety, transparency, and human impact in autonomous and learning-enabled control.

View on GitHub

Blogs

Contact

If you’d like to get in touch and seek my expertise, then feel free to get touch via the following platforms. I’m open to collaborations, mentorship and interesting projects.

GitHub