Open to Summer 2026 Internships

Vaibhav Shende

Robotics - M.Eng. at University of Maryland, College Park

Second-year M.Eng. Robotics student at UMD with hands-on experience in motion planning, perception, and control. Before grad school I spent a year building embedded robotics systems - C++ drivers, sensor integration, and SLAM pipelines on real hardware.

View Projects Resume
Vaibhav Shende

Projects

Things I've built or worked on — research projects, course projects, and personal experiments.

Project Zoomlet -- Multi-Robot ROS 2 Simulation

Project Zoomlet -- Multi-Robot ROS 2 Simulation

A complete differential drive robot stack built on ROS 2 and ros2_control. Supports spawning multiple namespaced robots in Gazebo Harmonic, joystick teleoperation with a deadman switch, and a full URDF/xacro model with physics-tuned parameters. Works on both Humble and Jazzy.

ROS 2 C++ Gazebo ros2_control URDF Diff Drive
View Source
Bug Path Planner -- Interactive Navigation Visualizer

Bug Path Planner -- Interactive Navigation Visualizer

An interactive grid-based visualizer for Bug0, Bug1, and Bug2 path-planning algorithms built with C++17 and SDL2. Paint obstacle maps, set start/goal/checkpoints, and watch the robot navigate in real time. Supports step-through, pause, speed control, and saving/loading maps.

C++17 SDL2 CMake Path Planning Bug Algorithm
View Source
Catch-a-Turtle -- ROS 2 Autonomous Chaser

Catch-a-Turtle -- ROS 2 Autonomous Chaser

A two-node ROS 2 system built on Turtlesim. One node spawns turtles at random positions and tracks which are alive; the other continuously finds the nearest turtle, drives toward it, and removes it on catch. Implemented in C++ with custom interfaces.

ROS 2 C++ Turtlesim Custom Interfaces
View Source

Perception-Driven Autonomous Navigation

Multi-capability perception stack on TurtleBot3: lane following, stop-sign detection with YOLOv8n, dynamic obstacle detection via optical flow, and horizon line estimation for navigation stability. Validated in Gazebo simulation and deployed on physical hardware with real-time inference.

ROS 2 Python YOLOv8 OpenCV Optical Flow Gazebo
Robot Motion Planning -- BFS / Dijkstra / A*

Robot Motion Planning -- BFS / Dijkstra / A*

Implemented classical planning algorithms (BFS, Dijkstra, A*) in Python applied to grid-based and continuous configuration-space environments. A* with Manhattan distance heuristic finds optimal solutions significantly faster than uninformed methods on harder configurations.

Python BFS Dijkstra A* Pygame Path Planning
View Source

Experience

Research and industry roles in robotics and embedded systems.

GAMMA Lab, University of Maryland

Active

Graduate Research Assistant — Advisor: Prof. Dinesh Manocha

College Park, MD

Oct. 2025 – Present
  • ▸ Contributing to CHOP (Counterfactual Human Preference Labels for Obstacle Avoidance), a visuomotor policy framework that reduced real-world collisions by 45.7% and improved goal success by 24.4% on the Ghost Robotics Vision60 quadruped.
  • ▸ Surveyed and analyzed state-of-the-art Vision-Language-Action (VLA) methods benchmarked in the paper, building a comparative understanding of visuomotor policy architectures and their safety limitations.
  • ▸ Constructed the counterfactual preference training dataset by designing the data-collection pipeline, pairing robot trajectories with counterfactual alternatives, and organizing human annotator labels for pairwise preference fine-tuning.
  • ▸ Implemented and validated CHOP experiments on the Ghost Robotics Vision60 quadruped, deploying the fine-tuned policy in real-world navigation scenarios and benchmarking obstacle clearance and path completion metrics.

9thPixel Geosoft

Robotics Software Engineer Intern

Hyderabad, India

Sep. 2024 – Nov. 2024
  • ▸ Integrated RTAB-Map and LiDAR-SLAM on the Unitree GO-2 quadruped, enabling real-time autonomous navigation and dense 3D mapping in unstructured indoor environments.
  • ▸ Built a wireless ROS 2 communication pipeline between the GO-2 and a host workstation, enabling low-latency telemetry and remote command execution over an unreliable link.
  • ▸ Architected the preliminary system design for a solar panel cleaning robot — defined sensor placement, motion constraints, and safety interlocks to prevent panel damage during autonomous traversal.

Octobotics Tech

Robotics Software Developer

Noida, India

Sep. 2022 – Aug. 2023
  • ▸ Contributed to an autonomous magnetic-wheel crawler robot designed to adhere to vertical ship hull surfaces and perform non-destructive testing (NDT) — including surface buffing and ultrasonic thickness measurement.
  • ▸ Built a C++ RS485 communication library for static torque sensors, providing real-time force feedback that allowed the crawler to maintain safe buffing pressure against hull surfaces without causing damage.
  • ▸ Designed a robust C++ actuator control stack over RS485 for smart motors, ensuring precise and reliable command execution critical for stable locomotion on vertical metal surfaces.
  • ▸ Integrated torque, IMU, and inspection modules with STM32 microcontrollers and bridged them to an NVIDIA Jetson Xavier via ROS (rosserial) for high-level autonomy processing.
  • ▸ Engineered fault-tolerant ground-station-to-robot communication using watchdog timers and heartbeat protocols, guaranteeing safe crawler shutdown under any connectivity failure.
  • ▸ Led development of a Qt5 ground-station GUI for full teleoperation of robot locomotion, 2-axis gimbal, and inspection peripherals — forming the operational foundation for future autonomous hull inspection.
  • ▸ Contributed to PCB design (KiCad), handling power distribution and signal isolation between embedded systems and high-voltage inspection modules.

Pashaan Technologies

Research Intern

Nagpur, India

Nov. 2020 – May 2021
  • ▸ Implemented an OpenCV-based vision system on an autonomous VTOL multi-copter to detect ArUco markers for precise autonomous landing maneuvers.
  • ▸ Designed a configurable Li-ion battery management system supporting series/parallel configurations to meet varying power requirements across platforms.
  • ▸ Built and calibrated a Core-XY 3D printer from scratch, enabling rapid hardware prototyping and faster iteration for the engineering team.
  • ▸ Built a software pipeline to parse and process raw SRT subtitle files, significantly reducing video post-processing time and resource usage.

Education

M

University of Maryland, College Park

College Park, MD

Master of Engineering in Robotics

GPA: 3.83 / 4.00
Jan. 2025 – Dec. 2026

Relevant Coursework

Motion Planning Robot Perception Autonomous Vehicles SLAM Machine Learning for Robotics System Design for Robotics

Skills

Tools and technologies I work with across robotics, embedded systems, and software.

Programming

C
C
C++
C++
Python
Python
MATLAB
MATLAB
CMake
CMake
Bash
Bash

Robotics & Simulation

ROS
ROS
NAV2
Nav2
GZB
Gazebo
R2C
ros2_ctrl
MVT
MoveIt
SLAM
SLAM
OpenCV
OpenCV
SF
Sen.Interface&Fusion

Hardware & Embedded

Arduino
Arduino
ESP32
ESP32
STM32
STM32
Raspberry Pi
RPi
JSN
Jetson
ZED
ZED
OUS
Ouster
GO2
GO-2

Protocols

485
RS485
UART
UART
I2C
I2C
SPI
SPI
CAN
CAN

Tools

SW
SolidWorks
F360
Fusion360
FCAD
FreeCAD
KiCad
KiCad
PRO
Proteus
Qt
Qt
Docker
Docker
Git
Git
Linux
Linux

Currently Learning

CUDA
CUDA
Docker Swarm
Swarm
ISAAC
Isaac Sim
CARLA
CARLA

Publications

CHOP: Counterfactual Human Preference Labels Improve Obstacle Avoidance in Visuomotor Navigation Policies
CHOP: Counterfactual Human Preference Labels Improve Obstacle Avoidance in Visuomotor Navigation Policies
Under Review 2026

CHOP: Counterfactual Human Preference Labels Improve Obstacle Avoidance in Visuomotor Navigation Policies

S. Gershom, A. Jianyu, S. Vaibhav, E. Sahire, A. Yaxita, M. Kondapi, C. Samarth, K. Jonathan Deepak, M. Dinesh

Robotics: Science and Systems

arXiv Video

Contact

Actively looking for a Summer 2026 internship in robotics, autonomy, or embedded systems. If you have something that might be a good fit, or just want to talk robots, reach out.

Send a Message

Find me at

I usually reply within a day or two.

Download Resume (PDF)