C++/Python/ROS/Machine Learning/Reinforcement Learning

Chinmay Burgul

*** ****** **, ********, ** Ó 508-***-**** R adfufd@r.postjobfree.com https://github.com/cmburgul Objective

Looking for full-time roles in the field of Robotics and Machine Learning Education

Worcester Polytechnic Institute (WPI) Aug ’18 – Ongoing Masters of Science in Robotics Engineering GPA: 3.84/4.0 SRM University Aug ’12 – May ’16

Bachelor of Technology in Mechatronics Engineering CGPA: 8.73/10 Technical Strengths

Programming languages : C++, Python, MATLAB

Software & Libraries : SQL, ROS, TensorFlow, Pytorch, OpenCV, PCL, Catia, Hadoop ecosystem.

Related Skills : Motion Planning, Reinforcement Learning, Deep Learning, Dynamics, Navigation & Control Work Experience

Research Assistant – WPI, Worcester, MA June ’20

Building a framework for increasing the grasp synthesis in robotic manipulation using Action Vision and Reinforcement Learning techniques.

Machine Learning Intern – Phood LLC, Boston, MA May ’19 – Aug ’19

Built a working setup of an image classification framework for food items.

Worked on pre-trained neural network models, Tensorflow, Cloud Clusters and AWS tools. ROS Ambassador – The Construct Jan ’18 – Aug ’18

Tutoring for ROS learning for motion planning and navigation packages. Technical Engineer – Entra Mechatronics Pte Ltd July ’16 – Mar ’17

Involved in developing food automation products, my roles included designing hardware by reverse engineering, strategic procurement of off the shelf components, vendor & OEM management Projects

Deep Learning and Computer Vision – WPI

Worked on Autoregressive models-PixelCNN, Autoencoders, VAE (VLAE, VQ-VAE, Pixel VAE, beta-VAE)

Worked on CapsuleNets, GAN’s (BiGAN, CycleGAN)

Explored types of skip-connection models (ResNet, DenseNet) and came up with a new architecture and compared its performance on CIFAR-10 dataset for image classification. Motion Planning using Reinforcement Learning Techniques – WPI

Path Planning in With In-hand Manipulation task on Friction finger setup with model free Reinforcement Learning Algorithms as DQN, DDPG + HER, PPO2.

Worked on Multi-goal, Multi-start Multi goal problems to perform versatile within hand manipulation and created a custom simulation environment.

Trained a policy to perform a dynamic task of pushing a puck on 6 DOF fetch robot. Worked on Mujoco, Openai gym environment & openai stable baselines.

Active Vision for Manipulation using Reinforcement Learning – WPI

Working on improving grasp synthesis for an unknown objects by training a viewpoint optimization RL policy to guide the direction of exploration.

Developing a dynamic RL environment to train a policy from the point cloud inputs. Working on ROS, C++, python, PCL and pytorch.

Motion Planning – WPI Feb ’19

Developed a planning framework for multi-agent scenario with centralized planning using search based algorithms and visualized the paths of agents in Rviz.

Developed a route planning framework for electric vehicles. Integrated and compared different search based algorithms and analyzed its complexity.

Sensor Fusion Ongoing

Developed a framework for lidar obstacle detection for autonomous vehicles by filtering, segmentation and clustering the point cloud data.

Developed a framework to calculate time to collision(TTC) by 2D based Feature tracking between camera images and fusion of camera image & Lidar data.

Contact this candidate