IISER Bhopal • Graduating May 2027 • CPI: 8.87 / 10.0

Shlok Mehndiratta

BS in Electrical Engineering & Computer Science (EECS)

Shlok Mehndiratta

IISER Bhopal

I am an undergraduate in EECS at IISER Bhopal, with a foundation in control theory and learning-based methods. Currently, my work spans aerial robotics, multi-robot coordination, and dexterous manipulation, focusing on how control algorithms interface with physical hardware.

My primary interests lie in the intersection of control systems, physical dynamics, and simulation environments. I enjoy working on how learning-based approaches can complement low-level controllers, enabling robots to handle contact, uncertainty, and complex interactions. Looking forward, I aim to pursue graduate study (MS/PhD) to work on pioneering research in robotics, dedicated to doing meaningful work that bridges scientific discovery with deep-tech innovation.

GitHub LinkedIn Scholar X

shlokm23@iiserb.ac.in Curriculum Vitae

01 Publications

Publications

Research papers in multi-robot swarms, area coverage, and state tracking.

All publications & BibTeX →

Under Review • ICRA 2027 2026

Multi-Robot Area Balancing In the Presence of Discrete Local Communication

Santosh Tantravahi, Nirman Jaiswal, Shlok Mehndiratta, P. B. Sujit, Aditya Paranjape

IEEE International Conference on Robotics and Automation (ICRA 2027)

Key takeaway: Switches dynamically between a greedy rule and an altruistic heuristic for fast convergence to equitable area partition under discrete, opportunistic communication; demonstrated in SITL simulation of multi-drone wildfire containment.

Swarm Robotics Discrete Communication Area Balancing

Accepted • ACODS 2026 2026

COAST: Altitude as a Coupled Perception Variable in Shared-Camera Aerial-Ground Tracking

Shlok Mehndiratta, Divyam Sood, Prashant Gupta, Agniva Banerjee, Arijit Sen

Advances in Control and Optimization of Dynamical Systems (ACODS 2026)

Key takeaway: Formulates the fundamental coupling between UAV altitude, measurement noise covariance (R ∝ z²), and camera footprint; proves estimation-optimal altitude scheduling under path curvature variations with 25× compute reduction via event triggering.

Perception-Aware Control UAV-UGV Tracking Belief-Space MPC

02 Technical Skills

Technical Skills

Technical proficiencies in programming, robotics & simulation, computer vision & ML, and systems deployment.

Curriculum Vitae →

Robotics & Control

ROS 2 (C++ / Python) • ros2_control • Model Predictive Control (MPC) • State Estimation (EKF) • Gazebo Harmonic • OptiTrack Mocap

Simulation & Synthetic Data

NVIDIA Isaac Sim • NVIDIA Warp (GPU Physics) • HDF5 Data Pipelines

Vision & Machine Learning

PyTorch • TensorFlow • OpenCV • Visual SLAM / Odometry • Reinforcement Learning

Scientific Computing & ML

NumPy • Pandas • scikit-learn • MATLAB / Simulink

Languages & Core Systems

Python • C++ (17/20) • Verilog HDL • Linux (Bash / POSIX) • Docker • Git / GitHub CI • LaTeX

03 Research Experience

Research Experience

Current and past research across aerial robotics, manipulation, and state estimation.

Full research timeline →

Robotics Research Center (RRC), IIIT Hyderabad

Advisor: Dr. Spandan Roy

Research Intern April 2026 – Present

Synthetic Data & Grasp Estimation for Dexterous Robot Hands

Working on synthetic simulation and vision-based grasp estimation for the 16-DoF LEAP Hand. Because physical robot hand experiments are slow to collect, we generate synthetic training data with camera images and contact maps to train grasping policies.

Technical breakdown

• Synthetic Data Generation: Built an automated pipeline generating synthetic multi-view RGB-D images and contact heatmaps for multi-object scenes, saved into structured HDF5 files.
• Physics & Sensor Simulation: Tested stable hand grasps in simulation, rendering scenes with path-traced lighting and sensor noise to closely match real camera feeds.
• Pipeline Batching & Caching: Improved rendering throughput across GPUs by organizing batch scenes and caching 3D asset files in memory to avoid repeated loading.
• Grasp Affordance Learning: Trained neural models on simulated data to predict hand approach poses and contact points for grasping objects in cluttered scenes.

Python PyTorch Robot Manipulation 16-DoF LEAP Hand Simulation HDF5

MOON Lab, IISER Bhopal

Advisor: Dr. P. B. Sujit

Undergraduate Research Student May 2026 – Present

Vision-Based Autonomous Navigation and Swarm Coordination

Developing vision-based trajectory planning and decentralized multi-drone coordination algorithms, transferring flight policies from ROS 2 simulation onto physical Crazyflie quadrotors in an OptiTrack motion capture arena.

Technical breakdown

• Vision-Based Navigation: Designing trajectory planning algorithms for aerial vehicles that actively respect camera field-of-view limits during dynamic obstacle avoidance and target tracking.
• Hardware Assembly & Integration: Assembled Crazyflie micro quadrotors with custom radio hardware and configured OptiTrack optical motion capture with ROS 2 for millimeter localization.
• Swarm Coordination: Investigating decentralized algorithms for multi-drone teams to achieve cooperative area coverage and balanced patrol duties under intermittent communication.

ROS 2 Crazyflie Swarms OptiTrack Mocap Trajectory Planning C++ Python

MuSic Lab, IISER Bhopal

Advisor: Dr. Arijit Sen

Undergraduate Research Student Sept 2025 – July 2026

COAST: Perception-Aware Aerial-Ground Tracking

In GPS-denied environments where a single downward-facing UAV camera serves as the ground robot’s only sensor, flight altitude creates a fundamental trade-off: flying higher expands lane coverage but degrades image resolution and measurement accuracy. We treated flight altitude as an active control variable and analytically derived an optimal altitude policy to balance this coverage-accuracy tradeoff, developing COAST (Coupled-Altitude Shared-camera Tracking).

Technical breakdown

• Optimal Altitude Policy: Modeled how camera measurement noise increases with flight altitude, proving the estimation-optimal altitude equals the lowest height that keeps the track search region inside the camera field-of-view.
• Belief-Space Control & Event-Triggering: Built a belief-space Model Predictive Controller (MPC) for the ground vehicle and an event-triggered aerial planner that only recalculates trajectories when tracking uncertainty requires it, cutting computation by 25× without losing accuracy.
• Full-System Gazebo Validation: Implemented modular ROS 2 nodes for lane segmentation, ArUco fiducial pose estimation, and Extended Kalman Filter (EKF) sensor fusion, validating the altitude scaling laws across curved tracks.

COAST Open-Source Code (GitHub) ↗

ROS 2 Gazebo Model Predictive Control (MPC) Extended Kalman Filter (EKF) Computer Vision Python

04 Open Source

Open Source Contributions

Contributions to the Robot Operating System 2 (ROS 2) control framework.

View open source contributions →

ros-controls • Contributor

ros2_control: Robot Operating System 2 Control Framework

Aug 2024 – Present

Contributed to ROS 2's real-time control framework by migrating controller locks to priority-inheritance mutexes to prevent latency spikes in high-frequency hardware loops, along with test suites for controller error handling.

C++20 ROS 2 Real-Time Control Thread Safety POSIX RT Mutex

5 Upstream PRs ↗

05 Selected Projects

Selected Projects

Engineering builds across computer vision, deep learning, and digital hardware.

All projects & coursework →

Robotics & Control Systems • Linear Control Systems (ECS 334) • IISER Bhopal

Robust Linear Control: A Belief-Coupled Tube-MPC Framework

2025 – 2026

Developed a unified research framework for autonomous tracking that bridges robust control theory, high-dynamic flight trajectory validation, and closed-loop ROS 2 / Gazebo simulation.

Details & architecture

• Derived tracking feedback gains and tightened nominal state/input constraints using Robust Positively Invariant (RPI) sets under bounded external disturbances.
• Validated high-dynamic 3D flight trajectories for micro-UAV hardware using Crazyflie SITL simulation containerized with Docker.
• Integrated perception (OpenCV), state estimation (Extended Kalman Filter), and belief-space control within modular ROS 2 nodes in Gazebo Harmonic.

Outcome: Guaranteed trajectory tracking stability within bounded disturbance tubes while closing the loop between perception, state estimation, and predictive control.

ROS 2 Jazzy Gazebo Harmonic Tube-MPC Robust Control (RPI Sets) EKF Crazyflie SITL Docker

Repository ↗

Computer Vision & 3D Learning • GeoAI4Cities Lab (Advisor: Dr. Vaibhav Kumar)

3D Point Cloud Segmentation with Graph Neural Networks

Dec 2024 – Jan 2025

Built a machine learning pipeline that models 3D LiDAR point clouds as graphs connecting points to geometric neighbors, using RANSAC filtering to classify urban scenes.

Details & architecture

• Represented 3D point clouds as localized graphs where each point connects to its nearest geometric neighbors to capture local topological shape.
• Applied RANSAC geometric filtering to suppress sensor reflections and sparse LiDAR noise before neural inference.
• Trained 3D deep graph networks in PyTorch, outperforming point-by-point classifiers in complex urban scenes.

Outcome: Showed that learning from local geometric neighbors produces cleaner 3D boundaries than treating points in isolation.

PyTorch Open3D RANSAC Python 3D Vision

Embedded Systems & Hardware • IIT Hyderabad (with Dr. G V V Sharma) / IISER Bhopal

Digital Logic Design & Bare-Metal AVR Microcontroller

Oct – Nov 2025

Designed fundamental digital logic circuits on breadboards and programmed an ATmega328P microcontroller using bare-metal AVR assembly and direct register manipulation, without external runtime libraries.

Details & architecture

• Wrote AVR assembly from scratch, directly writing to hardware control registers to manage timers, pin interrupts, and I/O ports.
• Designed synchronous digital circuits including decade counters, decoders, and discrete logic gates.
• Wired physical breadboard circuits to drive multiplexed seven-segment displays with cycle-accurate timing.

Outcome: Built a concrete, machine-level foundation in microcontroller execution, memory-mapped I/O, and synchronous digital logic.

AVR Assembly Direct Register I/O ATmega328P Digital Logic Circuit Design

Repository ↗

06 Notebook & Blogs

Notebook & Blogs

Personal notes, workshop reflections, and engineering logs from IISER Bhopal.

All notes (3) →

Feb 8, 2026 • 5 min read

Organizing an ESP32 IoT Workshop at IISER Bhopal

How we organized a hands-on hardware workshop with the EECS Club at IISER Bhopal, prepared setups in the hostel common room, and mentored junior undergraduates.

Read →

Side-Channel Timing Attack on FPGA Dev Kit

Nov 20, 2025 • 5 min read

Timing Side-Channel Analysis on an FPGA: A Digital Circuits Course Project

How an open-ended course project for Digital Circuits and Systems at IISER Bhopal turned into an experiment on timing leaks, legacy hardware toolchains, and designing secure logic.

Read →

07 Gallery