Shashank Rao Marpally

Shashank Rao Marpally

PhD Student (cleared Qualifying Examination)

National University of Singapore

About

Hi! I am a Second year Ph.D. Student at the National University of Singapore, currently working with Dr. Harold Soh at the Collaborative, Learning, and Adaptive Robotics (CLeAR). My research focusses on Robot Learning, Human-Robot Collaboration and Classical Planning. Previously, I worked on Learning action models of black-box agents through agent-interrogation with Dr. Siddharth Srivastava and before that, I worked on AI agents that generate explanations for planning tasks for humans with Dr. Yu (Tony) Zhang. When I’m not working on research, you would generally find me either playing some guitar, or cooking, or sketching, or watching british panel/comedy shows. I am also the Social Secretary for the NUS Graduate Student Society, where we organize career development as well as social events for the Graduate students of NUS.

UPDATE: I’m actively looking for research internship opportunities for 2023 Summer!

Interests

  • Robot Learning
  • Human Robot Interaction
  • Classical Planning

Education

  • Ph.D. in Computer Science, 2026

    National University of Singapore

  • Masters in Robotics and Autonomous Systems(Artificial Intelligence Concentration), 2021

    Arizona State University

  • B.Tech in Mechanical Engineering, 2019

    National Institute of Technology, Karnataka

Experience

 
 
 
 
 

Ph.D. Student

CLeAR Lab, National University of Singapore

Aug 2021 – Present Singapore
I’ve cleared my Ph.D. Qualifying Examination! I currently work in the areas of robot learning, human-robot interaction and classical planning.
 
 
 
 
 

Robotics Intern

Toyota Material Handling

Jun 2020 – Jul 2020 Indianapolis
Software Development for Autonomous Forklift Simulation
 
 
 
 
 

Research Assistant

Arizona State University

Jan 2020 – May 2021 Tempe,Arizona
Researching Action Model Learning of Black Box Agents (GPA: 4.0/4.0)
 
 
 
 
 

Research Intern

Indian Institute of Technology, Kanpur

May 2018 – Jul 2018 Kanpur,India
Researched Recurrent Neural Networks for Motion Planning on simulated UR5 Robot arm
 
 
 
 
 

Software Development Intern

Systemantics India Pvt. Ltd

Dec 2017 – Dec 2017 Bangalore,India
Developed UX for an Android App for Controlling a Robot, Assited in design of Infeed Mechanism for a process automation task
 
 
 
 
 

Research Intern

Indian Institute of Technology, Bombay

May 2017 – Jul 2017 Mumbai,India
Implemented a Multi-Robot Decentralized Graph Exploration Algorithm on ROS-Gazebo framework
 
 
 
 
 

Student

National Institute of Technology

Aug 2015 – May 2019 Mangalore,India
B.Tech in Mechanical Engineering (GPA: 3.77/4.0)

Publications

(2022). Discovering User-Interpretable Capabilities of Black-Box Planning Agents. In Proceedings of the International Conference on Principles of Knowledge Representation and Reasoning, 2021.

PDF

(2021). Learning User-Interpretable Descriptions of Black-Box AI System Capabilities. In Proceedings of ICAPS 2021 Workshop on Knowledge Engineering for Planning and Scheduling.

PDF

(2021). Asking the Right Questions: Learning Interpretable Action Models Through Query Answering. In Proceedings of the AAAI Conference on Artificial Intelligence, 2021.

PDF

(2021). Order Matters: Generating Progressive Explanations for Planning Tasks in Human-Robot Teaming. 2021 IEEE International Conference on Robotics and Automation (ICRA).

PDF

(2019). Geometrical Mapping of an Initially Unknown Region by a Mobile Robot. In Proceedings of IEEE DISCOVER, Manipal, India, 2019.

PDF DOI

Projects

ABU-ROBOCON 2018

Designed, developed, prototyped, fabricated and assembled (as a team) two industry-level robots that were to play a cooperative game of shuttlecock throwing.

Deep Reinforcement Learning Nanodegree

Completed all projects within the Deep Reinforcement Learning Nanodegree Offered by Udacity

Multi-Robot Decentralized Graph Exploration

Implemented an Exploration algorithm on ROS-Gazebo platform that utilizes multiple robots communicating in a decentralized fashion to achieve complete coverage of an apriori unknown map

RNN for Motion Planning

Investigated the applicability and utility of Recurrent Neural Networks for Motion Planning on a simulated UR5 robot arm in a ROS-Gazebo framework.

Robot Snakes mimic Snakes from Video using Genetic Algorithms

Utilized Genetic Algorithms to make a snake robot move like a snake from video from nature