hello i'm Gokul Prabhakaran

I am interested in building robots/software teaching/mentorship playing sports

About Me.

I am an incoming PhD student at Purdue University, planning to do research in the area of motion planning.

Contact me for collaboration/mentorship.

View Resume

Education

My education journey!

Image 1

Purdue University

PhD, Computer Science (in progress)

Aug 2025 - Current
Image 1

University of Michigan

Master of Science - MSc, Robotics

Aug 2022 - May 2024
  • - Grade: 4.0/4.0
  • - Research Assistant at the Biped Lab under Professor Jessy Grizzle
  • - Research Assistant at Ozay's Group under Professor Necmiye Ozay
  • - Graduate Student Instructor (GSI) for Graduate level Robotic Systems Laboratory, ROB 550
  • - Tau Beta Pi Member
  • - Relevant Courses: Principles of Machine Learnig, Mobile Robotics, Applied Optimal Control Robot Learning and Control, Deep Learning for Robotics, Autonomous Vehicles
Image 2

National University of Singpaore

Bachelor of Engineering, Mechanical Engineering (Robotics)

Aug 2018 - May 2022
  • - Grade: 4.46/5.0
  • - Minor in Computer Science
  • - Dean’s List for AY2020/2021 Semester 2
  • - Reserve Tutor for Introduction to Software Engineering, CS2103
  • - Relevant Courses: - Robot Dynamics and Control, Feedback Control Systems, Robotic System Design, Introduction to Software Engineering, Discrete Mathematics, Introduction to AI

Work Experience

Valuable work in the industry!

Image 1

Dexmate

Robot Arm Manipulation Intern

Nov 2024 - June 2025
  • - Developed and deployed Motion Planning stack on a real bimanual 17 DOF robot
  • - Whole body motion planning, with self collision avoidance
  • - Created primitive collision meshes for faster collison checks
Image 1

Panasonic R&D Center Singapore

Robot Arm Manipulation Intern

May 2022 - July 2022
  • - Implemented a research paper to enable UR5 robot to grasp unseen objects from 3D point cloud
  • - Created a program to segment a table-top from the point-cloud data. Its pose and size were estimated and added to the Moveit! planning scene for collision avoidance
  • - Created a custom dataset of objects and trained a Yolov5-Oriented Bounding box model. This model together with depth information was used to estimate the 3D position and the yaw angle of known objects for grasping
  • - Setup a Simulation Environment of UR5 robot, Robotiq gripper and Realsense camera in Gazebo and Rviz for testing algorithms
Image 2

Advanced Remanufacturing and Technology Center (ARTC)

Mobile Manipulator and Navigation Intern

July 2020 - Nov 2020
  • - Developed a GUI to manipulate a Fetch robot arm, using the Moveit! package. Utilized SLAM for mapping and navigation of the robot
  • - Programmed a pick & place demo for a mobile manipulator. Generated a 3D octomap and did point cloud filtering for grasping objects, while avoiding collision
Image 2

SHADO Group

Robot Design Intern

May 2020 - July 2020
  • - Developed a modular functionality for existing mobile vehicles, to follow a person/target, using Infrared Sensors
  • - Took charge of the mechanical design, electronics set up and software programming of the whole system and produced a working prototype in 2 months
Image 2

Amaris AI

Artificial Intelligence Intern

May 2019 - July 2019
  • - Built a computer vision software to perform image processing on ID cards and university offer letters to extract important information that helped to automate manual registration and verification processes
  • - Developed a software that can automatically extract documents and information from emails for processing

Projects

Recent cool projects that I have done.

post-img
Current

Backward Reachability for multi-link systems @ Umich-Ozay lab

  1. - Validation of state-of-the-art optimal control toolboxes for Hamilton-Jacobi Reachability Analysis on multi-link systems.
  2. - Working on optimal control algorithms to scale and speed up the computation of robust backward reachable sets.
  3. - More coming soon!.
post-img
Sept 23

Fall Prediction for Bipedal Robots @ UMich-BipedLab

  1. - Created a dataset of the Digit robot’s response to abrupt, incipient and intermittent forces, both in simulation and hardware.
  2. - Maximized lead time for fall prediction while minimizing false positive rates, using CNN
  3. - Developed an algorithm capable of detecting lead time of the robot, when subjected to abrupt, incipient or intermittent forces.
  4. - Accepted to ICRA 2024.
post-img
Dec 23

Drone motion planning and trajectory tracking using MPC

  1. - Derived the equations of motion of a drone from first principles and simulated a drone model on MATLAB.
  2. - Implemented RRT* with minimum-snap trajectory to plan a path to a goal pose, around obstacles.
  3. - Designed a linear and a non-linear MPC controller using quadratic and non-linear programming techniques to optimize reference trajectory tracking
post-img
April 23

Learning dynamics using NeuralODEs and controlling via MPPI

  1. - Data of the robot arm pushing the block is collected.
  2. - A NeuralODE model was trained on the data to learn the dynamics of the block.
  3. - This learned dynamics is then used to control the block to the target pose using a MPPI controller.
post-img
Aug 23

Robot arm manipulation and Autonomous Mobile robot

  1. - Derived forward and inverse kinematics for a 5DOF robot arm from scratch and tested it on the hardware.
  2. - Integrated a Computer Vision system with the robot arm to identify and stack colored blocks of different sizes.
  3. - Implemented a PID controller for a mobile robot to follow linear and angular velocity commands.
  4. - Developed algorithms on the mobile robot for mapping and localization of the surroundings using a LiDAR.
  5. - Integrated a A* path planner and a frontier explorer onto the robot, to enable autonomous mapping and navigation of the surroundings.
post-img
May 23

Robot arm manipulation and Perception @ Panasonic

  1. - Implemented ideas from a research paper to enable UR5 robot arm to grasp unknown objects from 3D point cloud.
  2. - Created a program to segment a table-top from the point-cloud data. Its pose and size were estimated and added to the Moveit! planning scene for collision avoidance.
  3. - Created a custom dataset of objects and trained a Yolov5-Oriented Bounding box model. This model together with a depth camera was used to estimate the 3D position and the yaw angle of known objects for grasping .
  4. - Setup a Simulation of UR5 robot, Robotiq gripper and Realsense camera in Gazebo and Rviz for testing.
post-img
July 22

Mobile robot manipulation and Perception @ ARTC

  1. - Developed Programs to move the joints of the robot arm, using the Moveit! package and ROS control package. Utilized SLAM for mapping and navigation of the robot .
  2. - Programmed a pick & place demo for a mobile manipulator. Generated a 3D octomap and did point cloud filtering for grasping objects, while avoiding collision.

Contact

Email: prabhakaran.gokul@gmail.com

Home about Contact

2025 Gokul