About Me

Who Am I?

Hi I'm Avnish Gupta. Robotic enthusiast and actively working in the fields of Computer Vision, Controls, Deep Learning for Robotics. Engaged in development of algorithms for perception, manipulation, system integration and control of mobile robots and industrial manipulators.

Strong engineering professional with a Bachelor of Technology (BTech) in Mechanical Engineering from the Indian Institute of Technology, Mandi with a strong focus on the fundamentals of computer programming. Have hands-on experience of working on ABB Industrial Manipulator Robot, Omron TM 14 Cobot, Lidar, stereo cameras, depth sensors, Servo Motors, etc. Passionate about learning new technologies and stays abreast of developments in the field of robotics.

Reseach

Learn

Tinker

Develop

What I do?

Here are some of my expertise

Mathematics

Linear Algebra, Calculus, Differential Equation, Probability Theory

Manipulation

Grasping and Pick & Place Algorithm for industrial manipulator arms

Control

Controller for docking of mobile robot, PID, Control Algorithms for industrial manipulator arms

Deep Learning

Object Recognition, Semantic and Instance Segmentation, Grasp Pose Prediction

Robot Perception

Primitive Shape detection and segmentation in Pointclouds, Obstacle Classification in Lidar Data, SLAM

Computer Vision

RGB camera and depth sensor realtime data fusion and calibration, Marker Detection, Template Matching

System Integration

IO Integration,End to end ROS integration with Robot, System behaviour design and implementation

Robotics

Probalastic Robotics, Manipulator kinematics and dynamics

Networking

TCP/IP socket, Modbus, Serial

Cups of coffee
Projects
Research Papers
Months
My Specialty

My Skills

Have hands-on experince working with following tools and technologies

Python, C++

80%

Robot Operating System(ROS)

100%

Quick Prototyping Boards

60%

OpenCV, PCL Library

80%

Stage, Gazebo, ADAMS

80%

Solidworks, ANSYS, MATLAB

60%

Tensorflow, Keras

60%

Linux

80%
Education

Education

Graduated in October 2018 as Bachelor of Technology in Mechanical Engineering from Indian Institute of Technology(IIT) Mandi

  • CGPA: 8.03/10

Went for Semester exchange program at Technical University of Munich(TUM)

  • GPA : 9.11/10

Vidya Bhavan Public School

  • Percent: 87.2%

SICA School

  • GPA: 9.6/10
Experience

Work Experience

Robotics Engineer July 2018 - Present

Addverb Technologies Pvt. Ltd., Noida, India

  • Developing a Deep Learning algorithm for recognition and instance segmentation of primitive shapes in Point Cloud data. Explored Mask-RCNN Technique for mask generation of unique planes of boxes in cluttered bin.
  • Developed an automated script for annotation of mask of planes in images of cluttered bin using background subtraction technique and constructed a data-set of 10000 images using the script
  • 3D Bin-picking Project:
    • Lead an interdisciplinary team of six people developing 3D Vision Based Bin-Picking using ABB 1200 robot attached with a vacuum gripper and achieved a throughput of 8-9 picks per minute.
    • Demonstrated the fully functional solution at Indian Warehousing Show 2019 held at Pragati Maidan, Delhi.
    • Published research paper on the developed solution at ACRA 2019.
  • Side Docking of Autonomous Mobile Robot(AMR): Developed and conceptualized perception and control algorithms for side docking of AMR using 2D LiDAR such that it docks sideways with respect to the conveyor in warehouses for box loading and unloading operation.
  • Multi-Robot Parcel Sorting System: Contributed to the development of Hybrid (Combination of Centralized and Decentralized Approach) Multi-Robot Path Planning algorithm for 200 mobile robots in simulation using Stage, ROS and python and achieved the throughput of 8000 sorts per hour.

Industrial Internship, Unmanned Weighbridge Dec 2017 - February 2018

A.D Technologies, Indore, India

Designed an IoT based autonomous truck positioning and weight lodging system for warehouses. Developed electronic circuitry and interfaced sensors, mechanical positioning system and networking algorithm for database management.

Research Internship, Human Biped Motion Analysis July 2017-November 2018

Professor Rajesh Ghosh, IIT Mandi, India

Analyzed dynamics of running and walking motion of a Biped and modelled it as a twelve-degree of freedom system imitating the human leg using MATLAB and ADAMS.

Projects

Projects

Addverb Technologies, Noida

  • Developed an improved version of 3D Vision Based Bin-Picking for box picking with dual sensors (camera + depth sensor) for perception
  • Devised calibration technique for fusing the camera image and depth sensor data for RGB-D point cloud data registration
  • Developed vision algorithm in two components: Box Localization and Box Segmentation with computation time less than one second. Generated masks of planes of boxes using image processing of RGB images in Box Localization part. Created algorithm to fuse the mask with the depth data which was further processed to extract 6 DoF poses of segmented unique planes of boxes for grasping in Box Segmentation part
  • Devised automatic calibration technique and developed algorithms using SVD for hand-eye calibration between robot and depth sensor using Aruco Markers

Addverb Technologies, Noida

  • Independently developed a perception algorithm for unique plane detection and semantic segmentation in a point cloud data using Point Cloud Library, Python and Open3D
  • Designed a linear actuator system having LiDAR and developed algorithms for stitching scene of cluttered bin from multiple 2D laser scans to 3D point cloud in real-time
  • Built simulation of the whole project in ROS and Gazebo with the overall implementation of the workflow
  • Independently developed a Robot Manipulation and Control Algorithm for pick and place using vacuum gripper in Move-it ROS

Addverb Technologies, Noida

  • Developed reflector and shape based docking marker detection algorithm from 2D laser scan and extracted marker center pose with respect to robot for side docking
  • Developed differential drive control algorithm for autonomously navigating the robot to the extracted center marker pose by the perception algorithm. The algorithm computes the most feasible and shortest path for the robot to reach the desired goal

Addverb Technologies, Noida

  • Developed a AR glass based system for aiding the mechanical assembly process using AR and computer vision algorithms. It helps the user by visually adding the part in the feed at the exact position on the jig where it is to be assembled. The assembly process is further monitored in real-time and user gets acknowledged
  • Captured live camera frame from AR glass camera using Droid Cam app and transferred the stream to Unity to augment the parts that are to be placed at each step of mechanical assembly in the live feed
  • Simultaneously, a pick by vision system was implemented using Open-CV to indicate user of the right part to be picked and where to place in the jig
  • Developed an template matching algorithm using Open-CV for real-time monitoring. The algorithm matches the pre-saved templates of each step to every frame of live feed and acknowledges the user on perfect placement of the component in the jig and proceeds to next step
  • Built web application using Django for broadcasting the processed live camera frame on a ip address for visualizing on AR glass

Noida

  • Developed a Python Framework for training Neural Network with custom Hand Gestures
  • Implemented custom image collector for autonomously taking images from a webcam. Developed preprocessing algorithm for removing noise, resizing and converting RGB images to binary images with subsequent labeling and storing them on the disk
  • Developed neural network trained by input images provided by the image collector, successfully recognizes new images from the webcam with 95% accuracy

Major Technical Project, IIT Mandi

Mentor: Professor Arpan Gupta

  • Developed an empirical model by experimenting on a self-manufactured scaled down airplane model and developed its numerical model using FEM and MATLAB. Subsequently accounted for flutter and non-linear frequency response of actual system

IIT Mandi

  • Led a team of six students and successfully designed, fabricated and tested a machine that dispatches grains based on the input given by the user via a touch screen, along with printing the bill. We integrated an LCD Touch Screen, servo motor, load cell, thermal receipt printer on a Raspberry Pi-2 development board and implemented the weight feedback control algorithm

IIT Mandi

  • Built a robot which senses obstacles in its proximity and passes without striking them. Implemented using Arduino UNO, UV sensors and a L298N motor driver
  • Implemented PID Control Algorithm for controlling the speed of the motors

IIT Mandi

  • Developed an Arduino based robot using IR sensors and DC Motors programmed to follow a black line as its path. Implemented PID Control Algorithm for a smooth Motion.
My Work

Recent Work

Bin Picking

Click to watch video

100 49

Read

Research Papers

HTML5 Bootstrap Template by colorlib.com
Australasian Conference on Robotics and Automation (ACRA) 2019

A Low Cost Bin Picking Solution for E-Commerce Warehouse Fulfillment Centers Draft

Video Link

Get in Touch

Contact