Soham Banerjee
SYSTEM.BOOT0%
RESUME

SOHAM BANERJEE

Robotics & Control Engineer
RESUME VIEW PROJECTS
SCROLL DOWN

My Arsenal

Soham Banerjee

Hi! I'm Soham — a Robotics & Control Engineer with a mechanical engineering base and a real love for control theory and deep learning. I build smart robots and 3D sims, always chasing what excites me 🚀

Currently researching underactuated bipedal locomotion at IIT Bombay
5+
PROJECTS
4
PAPERS
9.5
CPI

Control Systems & Dynamics

System IdentificationSystem Modeling & DynamicsLinear & Nonlinear ControlMPCCTCFeedback & Feedforward ControldSPACE

Robotics & Autonomous Systems

Robot Kinematics & DynamicsMotion & Path PlanningTrajectory GenerationSLAM & LocalizationROS / ROS2Webots

AI, ML & Reinforcement Learning

Regression & ClassificationPCA & LDADeep Neural NetworksCNN & RNNTransformersPyTorchPyBullet

Embedded Systems & Hardware

Arduino Uno / Nano / GigaESP32STM32TeensyJetson NanoRaspberry Pi

Software & DevOps

C++LinuxGit / GitHubDockerKubernetesLaTeX

CAD & Engineering Simulation

SolidWorksFusion 360ANSYSCOMSOLPyroSim / FDS3D Printing

Live GitHub Statistics

···Repositories
···Contributions / yr
···Followers

Contribution activity — last 12 months

Core Capabilities

My Work

Autonomous Navigation in Fire Environments using Drone

Reinforcement Learning / Aerial Robotics

Autonomous navigation for a Crazyflie 2.1 micro-UAV in fire-affected indoor environments, using only local temperature feedback and no prior map of the thermal field. Realistic fire scenarios were simulated via PyroSim/FDS and analytical superposition models, then evaluated with Greedy Hill Climbing, Epsilon-Greedy, and UCB multi-armed bandit strategies across single-source, multi-source, and sink-distorted thermal landscapes.

ROS2WebotsThermal ModelingRLPyroSim / FDSMulti-Armed BanditsPython
Find out more →
Planar biped rig, front view
Planar biped rig, angled view with ankle actuator

Underactuated Bipedal Locomotion with Ankle Push-Off

Control Theory / IIT Bombay Research

Open- and closed-loop control for an underactuated planar bipedal robot with pneumatic ankle push-off actuation, investigating precise impulse control during double-support transition phases. Built a unified modeling, simulation, and experimental framework from scratch, evaluating PID and Model Predictive Control (MPC) across multiple initial configurations and step sizes on both simulation and hardware-in-the-loop dSPACE setups to ensure gait stability and robust step tracking.

MPCPIDdSPACE HILPneumatic ActuationTrajectory GenerationPython / MATLAB
Find out more →
ADDPG actor-critic network architecture diagram

Virtual-to-Real Mapless Navigation via Deep RL

Deep Reinforcement Learning / Sim-to-Real

An asynchronous actor-critic (ADDPG) framework for continuous control of differential-drive mobile robots, using a compressed 14-dimensional state of sparse laser readings, relative target position, and previous velocity commands to output smooth linear and angular velocity commands. Trained entirely in PyBullet simulation and transferred to a physical Kobuki-based TurtleBot without any real-world fine-tuning, achieving collision-free mapless navigation at roughly 100 Hz control frequency.

ADDPGPyTorchSim-to-RealContinuous ControlMobile RoboticsPython
Find out more →

Motion Planning & Coordination for Autonomous Vehicles

Path Planning / Mobile Robotics

Configuration-space (C-space) modeling for a nonholonomic differential-drive TurtleBot, treating every pose as a point in the SE(2) manifold and inflating workspace obstacles via Minkowski sums to build free space. Benchmarked deterministic A* grid search against sampling-based RRT for planner efficiency, path length, and obstacle clearance, then closed the sim-to-real gap on hardware using AMCL localization over laser scans and an adaptive Pure Pursuit controller to compensate for wheel slip and onboard processing latency.

A* / RRTAMCLROSConfiguration SpacePure PursuitPython
Find out more →

Receding Horizon Control for Crazyflie Navigation

Optimal Control / Trajectory Optimization

Formulated and implemented an open-loop 2D drone trajectory optimization framework with double-integrator dynamics, Runge-Kutta 4th Order (RK4) discretization, and a smooth Softplus-based non-convex obstacle penalty. Optimized via CasADi's IPOPT solver to balance control effort, jerk minimization, terminal goal-reaching, and collision avoidance, then validated the resulting trajectories on real quadrotor hardware.

Trajectory OptimizationCasADi / IPOPTRK4Non-Convex OptimizationSoftplus PenaltyROS / WebotsPython
Find out more →

Find out more projects on my GitHub page

GitHub

My Timeline

Experience

Graduate Student Researcher — INDUS Lab, IIT Bombay

May 2026 – Present
IIT Bombay · Full-time | Mumbai Metropolitan Region · On-site
  • Working on hardware design and development of in-house planar biped robots and Adaptive Model Predictive Control–based walking with an ankle push-off mechanism

Undergraduate Student Researcher — IIEST, Shibpur

Jan 2025 – Dec 2025
IIEST, Shibpur · Full-time | India · Hybrid
  • Designed and implemented Computed Torque Control (CTC) for 2-DOF and 6-DOF robotic manipulators
  • Formulated LMI-based regional pole placement conditions to systematically design controller gains with prescribed stability and damping constraints
  • Developed nonlinear dynamic models incorporating joint friction effects, and analyzed controller performance with and without friction
  • Conducted extensive numerical studies and simulations in MATLAB and Simulink to evaluate trajectory tracking, stability margins, and robustness

Research Intern — VEHC Lab, Dept. of AM&BE, IIT Madras

May 2024 – Aug 2025
Indian Institute of Technology Madras · Internship | Chennai, Tamil Nadu, India · Hybrid
  • Performed numerical simulations using Wolfram Mathematica, MATLAB, and Simulink for modeling and analysis of dynamic systems
  • Investigated the nonlinear dynamics of a vertical cantilever beam subjected to inclined base excitations at varying angles
  • Analyzed the system's response for vibration-based energy harvesting, exploring the feasibility of piezoelectric (PZT) patch–based power generation

Research Intern — Indira Gandhi Centre for Atomic Research (IGCAR)

May 2024 – Aug 2025
IGCAR, Kalpakkam · Internship | Remote
  • Analyzed leak-before-break (LBB) criteria for design of secondary coolant piping systems in nuclear power plants to improve structural integrity and safety margins
  • Simulated hydraulic transient (water hammering) phenomena in pipeline networks using Deltares WANDA, evaluating time-varying pressure fluctuations to assess system reliability
IIT Bombay Photo coming soon
M.Tech, Systems & Control Engineering
IIT Bombay · CPI 9.52 · 2027
IIEST Shibpur Photo coming soon
B.Tech, Mechanical Engineering
IIEST Shibpur · CPI 9.8 · 2025
Kagram A.N.M High School Photo coming soon
Intermediate (WBCHSE)
Kagram A.N.M High School · 95.20% · 2021
Kagram A.N.M High School Photo coming soon
Matriculation (WBBSE)
Kagram A.N.M High School · 92.57% · 2019

Achievements