About Me
Hello there! My name is Saurav Aryan. I am a research engineer. Jack of all trades multiple trades.
With 7 years of experience in academic research environment, I am skilled in robotics, machine learning, programming and virtual reality
Motivated to work on innovative projects that requires learning new tools and doing cutting edge-research.
Please explore my projects in portfolio and my resume and get in touch if you think we could be a match.
- Name:Saurav Aryan
- Nationality: Indian
- Based in: Switzerland
- Permit: B (Work Authorized)
- Mail:contact@sauravaryan.com
Skills
C/C++ & C#85%
Python95%
ML: Tensorflow, Pytorch70%
ROS85%
LiDAR, SLAM70%
KUKA robot, Allegro hand70%
Unity: ARCore, VR70%
AWS: EC2, Lambda, S3, APIGateway40%
Languages
Education
-
2013 - 2015
Ecole Polytechnique, Paris
MS in Mechanics -
2009 - 2013
IIT Bombay, India
B. Tech., Mechanical Engineering
Experience
-
2018 - 2021
EPFL, Switzerland
Research Assistant -
2018
Seagate Technology, USA
Summer Intern -
2015 - 2018
University of Minnesota, USA
Research Assistant
Portfolio
-
Brain (EEG) signals exhibits a special type of patterns, called as Error Related Potentials (ErRP), whenever the subject notices something erroneous event (accident, collision and other shock-inducing events). We decode the EEG signals with supervised learning to check for the ErRP and use it as reward in an inverse reinforcement learning (IRL) framework to correct the robot's trajectory.
My role in the project:
- Research on reinforcement learning framework for robot control using EEG signals as rewards
- Installation and testing of SLAM algorithm to work with the LiDAR system
- Integrating the EEG signal acquisition hardware with the robot's computer
- Conducting experiments with the subjects.
- Institute LASA, EPFL
- Keywords LiDAR, SLAM, EEG decoding, Reinforcement Learning
- Date 2019-2021
-
Video of wheelchair with EEG equipment attached for experiments. Taken from Youtube
-
Tactile feedback is essential for humans along with visual feedback for grasping objects in everyday life. In this project, we integrated tactile feedback to prosthetic hand in a shared control framework. The subjects can control the fingers of the robotic hand through muscle signals (EMG) decoded by supervised learning. The shared control architecture further adapts the finger joint positions based on the input from the pressure sensors on the hand.
My contributions to the project:
- Implementation of shared control architecture on the KUKA robot in C++
- Setting up customized pressure sensors and integrating data acquisition system
- Integration of robotic system with EMG system on ROS platform
- Conducting experiments with the subjects
- Institute LASA, EPFL
- Keywords ROS, EMG, Prosthetic hand, KUKA, Allegro hand
- Date 2018-2019
-
Image Background Removal using Neural Networks running on Unity Barracuda inference engine.
Note: The neural network needs the server to be running on my computer. Without the server, you may only see your own video without neural network processing. Please contact to schedule a demo.
- Software Unity 2021
- Keywords VR, AR, WebGL
- Date 2021
Get in Touch