Lokesh Krishna

I am a PhD student at the Dynamic Robotics and Control Laboratory advised by Quan Nguyen at the University of Southern California, Viterbi School of Engineering. I received my Bachelor's in Electronics Engineering from the Indian Institute of Technology (BHU) Varanasi.

In my past life, I was a student collaborator working with Nidhi Seethapathi at the Seethapathi Motor Control Lab, MIT on the development of a framework to study human locomotor adaptation. Prior to that, I was a guest researcher at the Movement Generation and Control Group, Max Planck Institute for Intelligent Systems, working on control strategies for agile and dynamic bipedal locomotion under the supervision of Majid Khadiv. I was also a student researcher at the Stochastic Robotics Lab, Indian Institute of Science, jointly advised by Shishir Kolathaya and Ayonga Hereid, where I worked towards the development of a lightweight control frameworks for robust proprioceptive bipedal walking.

Apart from my research activities, I also co-founded and led the student research group, RoboReG at IIT BHU. As the technical lead of this group, I have successfully mentored several student-driven research projects and teams in events/competitions of national significance.

Email  /  CV  /  LinkedIn  /  Google Scholar  /  Twitter  /  GitHub

profile photo
Research

My research focuses on extending the athletic intelligence in robots, thereby leveraging their motor skills to be on par with that found in the animal kingdom: the result of centuries of evolution. Though nature stands as an unparalleled motivation for us to mimic through machines, I realize its cruel mockery and envy its athletic supremacy every time I see the energy-exploding gallop of a cheetah or the agile flight maneuver of a soaring eagle or the sudden sprint by a raging ostrich. Decades of development in robot actuation, sensing, and computation have finally paved the way for such machines to be a reality. However, the algorithmic principles to attain such extreme performance in robots while solving diverse tasks continues to be the final piece of the puzzle, calling to develop techniques that show strong empirical results backed by unassailable mathematical proofs and an understanding through first principles. To this end, I am interested in adopting formal methods from reinforcement learning and optimization to realize extreme performance in hybrid underactuated robotic systems with an optimal balance of performance, reliability, and guarantees.

Publications
Learning Multimodal Bipedal Locomotion and Implicit Transitions: A Versatile Policy Approach
Lokesh Krishna, Quan Nguyen
Accepted at IROS 2023
arXiv / video
Linear Policies are Sufficient to Realize Robust Bipedal Walking on Challenging Terrains
Lokesh Krishna*, Guillermo Castillo*, Utkarsh Mishra, Ayonga Hereid, Shishir Kolathaya
Accepted at IEEE RA-L (journal) and ICRA 2022
IEEE Xplore / arXiv / video
Learning Linear Policies for Robust Bipedal Locomotion on Terrains with Varying Slopes
Lokesh Krishna, Utkarsh Mishra, Guillermo Castillo, Ayonga Hereid, Shishir Kolathaya
Accepted at IROS 2021
IEEE Xplore / arXiv / project page / video
Robust Quadrupedal Locomotion on Sloped Terrains: A Linear Policy Approach
Kartik Paigwar, Lokesh Krishna, Sashank Tirumala, Naman khetan, Aditya Sagi, Ashish Joglekar, Shalabh Bhatnagar, Ashitava Ghosal, Bharadwaj Amrutur, Shishir Kolathaya
Accepted at CoRL 2020
PMLR / arXiv / project page / GitHub / video / slides
Projects
A Framework for Learning Human Locomotor Adaptation under Rigid Body assumptions
Lokesh Krishna, Inseung Kang, Nidhi Seethapathi

Learning Agile Bipedal Locomotion Through Trajectory Driven
Multi Stage Learning

Lokesh Krishna , Miroslav Bogdanovic , Majid Khadiv , Ludovic Righetti

GitHub (request acess) / video
AADOpt: A Framework for Antenna Array Design and Synthesis through Optimisation
Lokesh Krishna, Prityush Chandra, Prajwal Nair, MK Meshram

GitHub / preprint

Intelligent Picking: An end to end solution for ware house automation
Yash Sahijwani, Raghav Soni, Ayush Kumar Shaw, Niranth Sai, Lokesh Krishna

National Finalists, Flipkart Grid 2.0 Robotics Challenge

Slides / GitHub / video
Jerbot: a biomimetic bipedal robot
Lokesh Krishna, Nishant Kumar, Niranth Sai

GitHub / video
TOWRpy: a simulation test bed for TOWR trajectories in Pybullet
Lokesh Krishna, Shishir Kolathaya

GitHub
Mini Projects

MiniCheetahEnv: A modular pybullet environment for mini-cheetah with an MPC controller
GitHub / Video


ArduinoMLP: A NN netowrk library for Arduino
GitHub / Report / Video


RaisimStoch2: a simulation environment for Stoch2 in RAISIM
GitHub


LudoBot: An autonmous ludo solving robot
GitHub / Video


GestureBot: A mobile manpualtion platform for gesture controlled teleoperation
GitHub / Video


GAMCA: Genetic Algorithm to solve Maze like Cellular Automata
GitHub

Service
cs188 Technical Head, 2020-21

Founder, 2020

Coordinator, Robotics Summer Camp 2021

Mentor, Robotics Summer Camp 2020

Panel Member, 2020-21

Mentor, 2019-20


I am a Roboticist, I dont make cool websites. He does.