Shivam Thukral

Shivam Thukral

Software Engineer - Robotics and Perception

Locus Robotics

Hi!

My primary research focus revolves around the exploration of deep learning algorithms and their integration into robotic systems to enhance their intelligence, physical consistency, and seamless interaction with human counterparts. As a Software Engineer at Locus Robotics, I actively contribute to the advancement of autonomous mobile robots by enabling them to perceive their surroundings and make intelligent decisions.

I have recently completed my Master of Science degree in Computer Science from the University of British Columbia, where I had the privilege of being supervised by Ian M. Mitchell. Prior to my enrollment at UBC, I served as a research fellow at TCS Research and Innovation Labs, where I made contributions to the automation of warehouse robotics under the guidance of Swagat Kumar and Rajesh Sinha.

Additionally, I hold a Bachelor’s degree in Computer Science from IIIT Delhi, where I had the opportunity to work under the supervision of Rahul Purandare, and I actively collaborated closely with P.B. Sujit during my tenure.

My pronouns are he/him.

Interests

  • Robotics (Perception + Planning)
  • Computer Vision
  • Deep Learning for Images and Pointclouds
  • Competitive Programming

Education

  • MSc in Computer Science, 2022

    University of British Columbia, Vancouver

  • BTech in Computer Science Engineering, 2017

    Indraprastha Institute of Information Technology, Delhi

Experience

 
 
 
 
 

Software Engineer - Robotics and Perception

Locus Robotics

April, 2022 – Present Vancouver
At Locus, I am a member of the Robot Perception Team. In the coming months, I will give an update on my projects.
 
 
 
 
 

Graduate Research Assitant

University of British Columbia

May, 2020 – February, 2022 Vancouver

We propose to augment smart wheelchair perception with the capability to identify potential docking locations in indoor scenes. ApproachFinder-CV is a computer vision pipeline that detects safe docking poses and estimates their desirability weight based on hand-selected geometric relationships and visibility. Although robust, this pipeline is computationally intensive. We leverage this vision pipeline to generate ground truth labels used to train an end-to-end differentiable neural net that is 15 times faster.

ApproachFinder-NN is a point-based method that draws motivation from Hough voting and uses deep point cloud features to vote for potential docking locations. Both approaches rely on just geometric information, making them invariant to image distortions. A large-scale indoor object detection dataset, SUN RGB-D, is used to design, train and evaluate the two pipelines.

Potential docking locations are encoded as a 3D temporal desirability cost map that can be integrated into any real-time path planner. As a proof of concept, we use a model predictive controller that consumes this 3D costmap with efficiently designed task-driven cost functions to share human intent. This wheelchair navigation controller outputs a nominal path that is safe, goal-oriented and jerk-free for wheelchair navigation.

 
 
 
 
 

Graduate Teaching Assistant

University of British Columbia

September, 2019 – December, 2021 Vancouver

Designed and graded questions for homework assignments, quizzes and examinations for the following courses:

 
 
 
 
 

Research Software Engineer, TCS Research and Innovation Labs

TATA Consultancy Services

August, 2017 – August, 2019 Noida

I was involved in multiple research projects for warehouse automation using industrial manipulators. Specifically, I have worked on 3D pose-estimation heterogenous sized boxes using pointclouds and motion of planning for Universal Robots using ROS.
Following is a list of selected projects that I worked on:

  • Long Distance Container (LDC) Packing Video
  • Chitrakar: Robotic System for Drawing Jordan Curve of Facial Portrait Video
  • Amazon Robotic Challenge Video

Please refer to my Curriculum Vitae for a detailed description of these projects.

 
 
 
 
 

Undergraduate Research Assitant

Indraprastha Institute of Information Technology (IIIT), Delhi

August, 2016 – December, 2016 Delhi
Developed an optimal path planning algorithm BugFlood inspired by the bug algorithm to quickly compute paths in an obstacle rich environment (or to report that no such path exists). In this approach, we simulate virtual bugs that upon sensing an obstacle splits into two bugs exploring the obstacle boundary in opposite directions, until the bugs find the goal in the line-of-sight. We compare the performance of our algorithm with different planners from Open Motion Planning Library (OMPL) and Visibility Graph methods. The results show that the proposed algorithm delivers lower cost paths compared to other planners with lower computational time and rapidly indicates if a path does not exist.

Projects

*
All AI ROS Verification Vision Thesis

PyTorch Vision Tutorials

Multiple tutorials covering how to implement vision­focused deep learning architectures in PyTorch with torchvision.

ApproachFinder-NN

Developed an end-to-end docking location detection network based on synergy of deep point set networks and Hough voting.

ApproachFinder-CV

Developed a real-time computer vision pipeline to find potential docking locations indoor environments for wheelchairs using point cloud data.

Wheelchair Navigation

Real-time wheelchair navigation with shared control using model predictive path integral (MPPI) controller.

Real-time Indoor Object Detection

Indoor object detection using Votenet for pointclouds captured from RGB-D cameras in ROS simulation.

Visual Servoing

Image-based visual servoing in eye-in-hand configuration for Universal Robot 5 using Microsoft Kinect V2 camera.

Modelling-Human-Behaviour-in-Chess

Developed a predictive model that can play chess like humans, with special focus on modelling amateur play.

Verifying DNN

Summarised 10 state-of-the-art approaches to verify DNN and developed a framework to test networks (eg ACAS Xu) on safety cases using SMT solvers.

3D Human Pose Estimation

Developed a CNN capable of obtaining a temporally consistent, full 3D skeletal human pose from a single RGB camera.

Sudoku SAT Solver

Converted Sudoko as a Boolean Satisfiability Problem to solve it through SAT

Text Detection and Recognition in Natural Scenes

Studied and summarised major approaches to perform text detection and recognition using deep learning techniques.

BugFlood

Developed an optimal path planning algorithm in obstacle rich environments. BugFlood unlike its predecessor uses a split and kill approach to advance in the environment. Performance of this algorithm was compared with different planners from Open Motion Planning Library (OMPL) and visibility graph methods.

Resolving Message Logic Dependencies in ROS

Developed a static analysis Clang based tool for ROS to reduce network latency and dropout rate by optimizing message size.

Characterization Tool for C/C++ codes

A clang based tool to find different types of statements in C/C++ code. This tool is used generate meta-data for a ROS package.

Publications

Papers, Workshops and Patents

Quickly discover relevant content by filtering publications.
Nishant Sharma, Shivam Thukral, Sandip Aine, P.B. Sujit (2018). A virtual bug planning technique for 2D robot path planning. Presented at 2018 Annual American Control Conference (ACC).

PDF Slides DOI

Aniruddha Singhal, Harshad Khadilkar, Venkat Raju Chintalapalli Patta, Deepak Raina, Venkatesh Srinivas Prasad, Shivam Thukral, Rajesh Sinha, Richa Verma (2019). System and Method for Autonomous Multi-bin Parcel Loading System.

Source Document

Aniruddha Singhal, Ayush Kumar, Shivam Thukral, Deepak Raina, Swagat Kumar (2020). Chitrakar: Robotic System for Drawing Jordan Curve of Facial Portrait. Workshop on Creativity and Robotics, International Conference on Social Robotics (ICSR), 2020.

PDF Video

Recent Talks

ApproachFinder: Real-time Perception of Potential Docking Locations for Smart Wheelchairs

Thesis Project Presentation
Dec 15, 2021 Vancouver, British Columbia
Shivam Thukral
Slides

3D Object Detection for Indoor Scenes

Compared different state-of-the-art 3D object detection networks
May 15, 2021 Vancouver, British Columbia
Shivam Thukral
Slides

Visual Servoing using Industrial Manipulator

Image-based visual servoing using industrial manipulators
Dec 8, 2020 Vancouver, British Columbia
Shivam Thukral
Slides

Uncalibrated image-based visual servoing

IBVS without knowing camera parameters.
Nov 27, 2020 Vancouver, British Columbia
Àngel Santamaria-Navarro, Juan Andrade-Cetto
Slides
See all talks

Contact