Meghdeep Jana
412-983-9258 ! [email protected] è linkedin.com/in/megh ð meghdeepj.weebly.com
Education
Carnegie Mellon University Pittsburgh, PA
Master of Science in Mechanical Engineering - Research, GPA: 4.0/4.0 May 2023
Research Advisor: Prof. Jean Oh, CMU Robotics Institute, School of Computer Science
Coursework: Planning and Decision-making in Robotics, Computer Vision, Statistical Techniques in Robotics
Indian Institute of Technology Guwahati Guwahati, India
Bachelor of Technology in Mechanical Engineering, GPA: 8.94/10.0 Jun 2019
Coursework: Robotics and Applications, Control Systems, Kinematics and Dynamics, Mechatronics and Automation
Experience
Outrider Technologies Inc. Golden, CO
Robotics Software Engineering Intern May 2022 - Aug 2022
• Developed a complete proof-of-concept simulation environment from scratch in C++ and Python as a replacement for
the existing Gazebo sim to test and evaluate full robotic manipulation operations; reduced compute requirement by
25% and allowed auto-testing integration to the CI/CD pipeline by containerizing it using Docker.
• Designed the simulator to manage mesh poses, color and depth camera sensor models, a simulated ROS interface to an
industrial robot arm, and a custom trajectory controller that mimics control modalities of the real-world system.
Peppermint Robots (Aubotz Labs Pvt. Ltd.) Pune, India
Senior Robotics Engineer Jun 2021 - Jul 2021
• Developed high-level planning system of an autonomous floor cleaning robot using BehaviourTree.CPP in ROS;
incorporated features such as obstacle avoidance, cleaning zone chaining and recovery behaviours. Enabled robots to
operate in more dynamic sites such as airports and malls, initiating the first overseas pilot project at UAE.
• Maintained git servers for version control and ported ROS based Software stack for integration to floor cleaning
robots; reduced deployment time of stable software stacks by 30%.
Robotics Engineer Jul 2020 - Jun 2021
• Developed 3D obstacle detection and prediction system using state-of-the-art deep learning techniques on multi-modal
depth and range sensor data fusion; secured safety clearance for deploying robots to customers and Peppermint
completed its first on-site deployment.
• Collaborated with another team member to redesign the ROS architecture and port the navigation module of the
software stack from Python to C++; Improved the controller frequency from 5Hz to 30Hz on an Nvidia Jetson Nano.
Autonomous Robots and Multi-agent Systems Lab, IIT Bombay Mumbai, India
Research Assistant Dec 2019 - Jun 2020
• Designed a novel deep reinforcement learning based path planning for multi-agent collaborative patrolling using
decentralized Deep Q-Networks implemented on PyTorch. Demonstrated scalability to large grid-like environments
while being adaptive against adversaries and changes in graph structure.
• Authored an article and published to peer-reviewed International Journal of Intelligent Robotics and Applications.
Research Projects
First Person Social Navigation Simulator for Autonomous Mobile Robots Pittsburgh,PA
Advisor: Prof. Jean Oh, Bot Intelligence Group, CMU RI Sep 2021 - Present
• Developed a social navigation dataset generation method: T2FPV, in Unity, for constructing high-fidelity first-person
view datasets from real-world pedestrian trajectories
• Developed a deep learning module, CoFE, that reduces the impact of realistic perception errors on downstream
prediction performance by more than 10%; Authored a conference paper and submitted our findings to IROS 2023
Multi-Waypoint Motion Planning in Dynamic Environments for Delivery Robots Pittsburgh,PA
Advisor: Prof. Maxim Likhachev, Instructor for 16782 Planning in Robotics, CMU RI Oct 2021 - Dec 2021
• Developed a planning pipeline to generate high-level plan to select order of waypoints using TSP, a low-level A* plan
to get the optimal path to waypoints, and an obstacle trajectory prediction system using constant velocity model
• The resulting algorithm proved capable of handling dynamic environments of complex geometry and complicated
obstacle movement while generating efficient plans.
Skills
Programming Languages: C/C++, Python, MATLAB
Software Libraries Tools: ROS, Gazebo, Pytorch, OpenCV, PCL, BehaviourTree.CPP, Tensorflow, Open3D, Docker
Hardware: Nvidia Jetson, STM32, Raspberry Pi, Arduino
Leadership
4i-Labs, IIT Guwahati Guwahati, India
Project Manager, Antahsagari: Underwater Remotely Operated Robot Jul 2017 - May 2018
• Managed and mentored a team of 5 undergraduate students to engineer an underwater remotely operated robot from
ideation to working model; achieved 2nd position at zonal level in 1 of 5 zones in India at National DRDO Robotics
and Unmanned Systems Exposition 2018.
