Hi there! I am Manan Arora

I am a Final Year undergrad at BITS Pilani Goa Campus majoring in Electronics and Communication Engineering. I am interested in Robotics and Reinforcement Learning.

Currently, I am working as a Research Assistant at Stoch Lab under the guidance of Prof Shishir Kolathaya on Locomotion in Legged Robots using Reinforcement Learning. Presently, I am trying to tran robust locomotion policies for quadrupeds using Offline Reinforcement Learning.

Previously, I interned at Swaayatt Robots where I worked closely with Mr Sanjeev Sharma(Founder and CEO, Swaayatt Robots) on Motion Planning using Reinforcement Learning for Autonomous Vehicles.

Before that, I was a Research Intern at MARMot Lab, National University of Singapore working under the guidance of Dr Guillaume Sartoretti and his research group on Foothold Planning using Reinforcement Learning for Legged Robots. Even before that, I was a Research Intern at CSIR-CEERI under the guidance of Dr Kaushal Kishore.

My aim is to induce animal like conciousness and reasoning in robots especially for locomotion and manipulation tasks.

I am actively looking for Research Positions for Fall 2024. Check out my CV to know more about my work experience.

Experience

Stoch Lab

Swaayatt Logo [January 2024-Present]
Research Assistant
Advisor - Prof Shishir Kolathaya

Swaayatt Robots

Swaayatt Logo [June 2023-December 2023]
Research Intern
Advisor - Mr Sanjeev Sharma

MARMot Lab

Swaayatt Logo [February 2023-June 2023]
Research Intern
Advisor - Dr Guillaume Sartoretti

BITS Pilani Goa Campus

Swaayatt Logo [September 2022-Present]
Undergraduate Researcher
Advisor - Dr Sarang Dhongdi

CSIR-CEERI

Swaayatt Logo [June 2022-September 2022]
Research Intern
Advisor - Dr Kaushal Kishore

Project Kratos

Swaayatt Logo [June 2021-February 2022]
Team Member

 

Projects(ongoing)

Spiderbot image
  • Training a hexapod to walk on rough terrain, steps etc. using Motion Capture data and Deep Reinforcement Learning.
  • Motion Capture data generated using a custom built CPG controller.
  • Spiderbot image
  • Implemented an Atari like Reinforcement Learning environment and agent from scratch using Pygame.
  • Implemented PPO from scratch to train the agent.
  • Exploring training results from other algorithms like TD3, DDPG.
  •  

    Projects(selected)

    Lander img
  • Implemented paper "DreamWaQ: Learning Robust Quadrupedal Locomotion With Implicit Terrain Imagination via Deep Reinforcement Learning" from scratch using pytorch and IsaacGym.
  • Lander img
  • Implemented Batch Constrained Deep Q-Learning Algorithm (Fujimoto et al.) from scratch and tested on d4rl dataset.
  • Included modifications to speedup training by approximately 75%.
  • Lander img
  • Implemented PPO algorithm from scratch and tested on platforms like LunarLander, Bipedal Walker
  • Included modifications like Generalised Advantage Estimation, Entropy Regularisation etc. to improve performance
  • Landing GIF
  • Worked on a visual servoing based controller and Open-CV pipeline to detect and land on the platform.
  • Worked on Deep Learning and Computer Vision based methods to detect the platform using a 3D Lidar.
  • Spiderbot image
  • Implemented various gait patterns using bio-inspired CPG controller.
  • Currently working on Reinforcement Learning based algorithms for locomotion.