Gagan Khandate

gagank at cs dot columbia dot edu

I am a PhD student in Computer Science at Columbia University advised by Matei Ciocarlie working on robot learning as a member of Robotic Manipulation and Mobility Lab. Enabling complex motor skill acquisition through the combined use of models, data and domain knowledge is the focus of my research.

From 2015 to 2017, I worked as a Research Engineer at Systemantics where I primarily developed controls, kinematics and motion-planning firmware for serial and parallel collaborative industrial manipulators.

I received my master's from Columbia University in Mechanical Engineering with a focus on Robotics and recieved my bachelor's from Indian Institute of Technology Madras also in Mechanical Engineering with a minor in micro-electronics where I was advised by Sandipan Bandyopadhay and C. Sujata

GitHub  /  Google Scholar  /  LinkedIn

profile photo


  • NEW! [2022-05] Excited to be spending summer at Amazon Robotics AI, Boston.
  • NEW! [2022-02] Paper on in-hand manipulation accepted to ICRA 2022
  • [2021-05] Invited talk on using tactile sensing for in-hand manipulation at ViTac Workshop in ICRA 2021
  • [2021-03] Guest lecture on Model Predictive Control in our course on Robot Learning
  • [2020-05] Paper on snake locomotion accepted to ICRA 2020
  • [2019-11] Lecture discussing a seminal paper on vision based grasping in our course on robot learning.


project image

On Feasibility of Learning Finger-gaiting In-hand Manipulation using Intrinsic Sensing

Gagan Khandate, Maximillian Haas-Heger, Matei Ciocarlie
IEEE International Conference on Robotics and Automation (ICRA), 2022
arxiv / video /

In this work, we use model-free reinforcement learning (RL) to learn finger-gaiting only via precision grasps and demonstrate finger-gaiting for rotation about an axis purely using on-board proprioceptive and tactile feedback. To tackle the inherent instability of precision grasping, we propose the use of initial state distributions that enable effective exploration of the state space.

project image

Automatic snake gait generation using model predictive control

Emily Hannigan, Bing Song, Gagan Khandate, Maxmillian Haas-Heger, Ji Yin, Matei Ciocarlie
IEEE International Conference on Robotics and Automation (ICRA), 2020
arxiv / video /

In this paper, we use Model Predictive Control (MPC) with iLQR to automatically generate effective locomotion gaits via trajectory optimization. An important advantage of our method is that it can be applied to generate gaits under different environment dynamics such as dry friction or viscous friction while also enabling obstacle avoidance.

project image

SBP-Guided MPC to Overcome Local Minima in Trajectory Planning

Emily Hannigan, Bing Song, Gagan Khandate, Maxmillian Haas-Heger, Ji Yin, Matei Ciocarlie
Optimal Planning and Control: Fusing Offline and Online Algorithms Workshop, ICRA, 2019
arxiv / video /

Non-linearity of robot dynamics can cause failure of trajectory optimization methods (ex. iLQG) due to local optima. On the other hand, Sampling Based Planning (SBP) methods such as Rapidly-exploring Random Trees (RRT) are inherently robust to presence of local optima but often generate in-efficient trajectories. We propose combining these two classes of methods to retain the strengths of each. We use RRT trajectory as initialization for iLQG and overcome local optima while producing an efficient trajectory.

project image

Algorithmic Gait Synthesis for a Snake Robot

Gagan Khandate, Emily Hannigan, Maximilian Haas-Heger, Bing Song, Ji Yin, Matei Ciocarlie
Toward Online Optimal Control of Dynamic Robots: From Algorithmic Advances to Field Applications Workshop, ICRA, 2019
arxiv /

In this work, we study the use of deep reinforcement learning for control of snake robots. While prior work on control of snake robots primarily uses open loop control backed by sinusoidal gaits (serpenoid curves), we demonstrate the use of deep reinforcement learning (PPO) for generating snake gaits under different environments. We also compare our method with other classes - model predictive control (MPC) and sampling based planning (SPB). Our results show that the gaits generated by model-free deep reinforcement learning are comparable (sometimes better) to MPC in terms of efficiency and energy comsumption.

Other Projects

These include coursework, side projects and unpublished research work.

project image

Safe RL for Islands & Bridges: A Topologically Isomorphoic Problem to In-hand Manipulation


Dexterous in-hand manipulation can be thought of traversing narrow regions of stable state space while moving between relatively more stable grasp poses. Mimicing this structure, we construct a topologically isomorphic problem, islands and bridges, wherein we connect large islands by narrow bridges to correspond to stable and unstable state-space respectively. Here, we demonstrate that using a shield to ensure safe actions signficantly improves sample complexity for goal based navigation. Our results suggest safe reinforcement learning for enabling model free learning for in-hand manipulation.

project image

roamcluster: docker-based job scheduler


A docker-based job scheduler with dynamic compute pool. It consists of a central database of jobs and a number of distributed machines that pick up jobs from the database.

project image

Hindsight in On-policy RL


In this work, we demonstrate an approach to use hindsight experience in on-policy RL methods for goal based problems using a implicit form of importance sampling. We demonsrate significant improvement in sample efficiency for problems with dense reward but more importantly enable learning with sparse rewards previously demonstrated with only off-policy methods.

project image

Neuroevolution for Navigation


In this work, we evolve neural network policies via mutation and recombination for dynamic obstacle avoidance. Traditonal approaches to mobile robot motion planning invovlves computationally expensive steps - mapping and planning. However, in this exploratory work, we simply map perceptions to actions using neural network policies and evolve them for dynamic obstacle avoidance.

Design and source code from Jon Barron's website