Control Theory, Optimization, Learning.
I work on the kinematics, dynamics, and control of robotic manipulators and aerial systems, from link length optimization of redundant arms to compliant manipulation on the Franka Emika Panda. I'm graduating from NIT Tiruchirappalli and heading into a Master's in Robotics Engineering at WPI, working toward a PhD in manipulation and whole body control.
Academic Background
Worcester Polytechnic Institute · Worcester, MA
Incoming. Coursework planned in robot dynamics, robot control, and deep learning for perception.
National Institute of Technology, Tiruchirappalli · CGPA 8/10
NITT Undergraduate STEM Research Fellowship Awardee.
Recognition
Top Merit Award · RECAL, The Alumni Association of NIT Trichy
Awarded a research grant of ₹70,000 for a proposal on adaptive reinforcement learning for predictive control of autonomous quadrotor systems.
View certificate ↗NPTEL · IIT Madras, Instructor: Dr. B. Ravindran
View certificate ↗NPTEL · IISc Bangalore, Instructor: Dr. Suresh Sundaram
View certificate ↗Research & Work
xTerra Robotics · IIT Kanpur
Bachelor's Thesis · NIT Trichy & xTerra Robotics · Guides: Prof. K. Srinivasan, Amritanshu Manu
NIT Tiruchirappalli · Guides: Prof. K. Srinivasan, Prof. T. Shuprajhaa
Bhatiyani Astute Intelligence Limited · Bengaluru
NIT Tiruchirappalli · Guides: Prof. Ramakalyan Ayyagari, Prof. Subbareddy Chitta
Indian Institute of Science, Bengaluru · Guide: Prof. Jishnu Keshavan
Research Output
2026 IEEE North East India International Energy Conversion Conference and Exhibition (NEIICON)
Read paper ↗5th International Conference on Emerging Electronics and Automation (ICEEA)
Read paper ↗9th International Conference on Inventive Systems and Control (ICISC)
View on IEEE Xplore ↗Selected Work
Implemented a 6-DOF Cartesian impedance controller for compliant manipulation, reaching 3.8 mm peak and 0.14 mm steady state tracking error. Built a two-camera 3D localization pipeline for the Franka Emika Panda in MuJoCo, achieving 1.4 mm position and 0.1 degree yaw accuracy without learned models. Closed the loop with a pick and place system combining wrist camera pose refinement, damped least squares IK, and contact triggered grasping, validated across 50 randomized trials with a 100% success rate.
Trained TD3 and DDPG actor critic agents to dynamically tune controller gains, with a reward function shaped on LQG cost. Demonstrated disturbance rejection under external perturbations and reached 60% lower cumulative tracking error than fixed gain PID, LQR, and LMPC baselines.
Derived a quasi linearized fourth order inner loop model from a nonlinear longitudinal missile autopilot using feedback linearization. Designed a Mamdani Type 1 fuzzy inference system in MATLAB for online adaptation of LQR weights, improving autopilot robustness by 28.92% over offline tuning.
Built the kinematic model with the Craig DH convention and optimized link lengths with NSGA-II. Computed actuator torques with the Recursive Newton-Euler Algorithm at a 5 kg payload. Solved inverse kinematics with a damped least squares solver using adaptive Gaussian damping, reaching 100% convergence.
Worked within a 15-member team on the conceptualization, analysis, and validation of the suspension and steering system for a 4WD all-terrain vehicle, from design through fabrication. Reduced front and rear track widths by 3.77% and 2.13% to improve off-road handling, contributing to a 19th place finish at mBAJA 2024, organized by SAE India.
Identified a discrete-time linear state-space model of a mobile robot's nonlinear dynamics with the N4SID algorithm, using recorded acceleration and control input data. Tuned a PID controller on the identified model with the Ziegler-Nichols method and validated closed-loop tracking and stability in Simulink.
Toolset
Get in Touch
Reaching out about manipulation research, control systems, or WPI? Happy to talk. Best by email or LinkedIn.