Home / Sahayak- An Autonomous COVID Bot

Sahayak- An Autonomous COVID Bot

Introduction

Medical personnel regularly put themselves at risk of infection to protect patients, a fact that everyone has been acutely aware of during the COVID-19 pandemic. Development of an autonomous robot for hospital environment aiming to reduce the risk of the health care workers at their work has exponentially increased in the ongoing pandemic.  Sahayak, a COVID aid bot that is capable of navigating autonomously in simulated static environments. The aim is to provide a general framework for developing medical assistive robots.  A prototype of Sahayak was successfully deployed at AIIMS Nagpur.  

Mechanical Design

  • To enable smooth motion, a 4-wheel drive design was adopted. Sahayak has a payload capacity of 20 Kg and is driven by four planetary DC geared motors.
  • It also includes heavy-duty disc wheels and 20Ah lithium-ion battery to power the motors.
  • The prototype of Sahayak included a Jetson Nano for computation.
  • A A2M8 2D LiDAR  and an Intel RealSense R200 was mounted on the model. 

Previous Work

S​​ahayak v1:

  • Sahayak v1 started as a simple 4 wheel drive which was teleoperated using a remote control

Sahayak v2:

  • The new version had improved motors, ability to carry over 20kg payload and other facilities such as Video Conferencing.
  • It was successfully deployed at AIIMS Nagpur. 

Pipeline Overview

Pipeline

Visual Odometry

  • Visual Odometry was computed through 3D-2D and 3D-3D motion estimation methods by using a sparse optical flow-based feature tracking approach. 
  • This odometry information was fused with the IMU readings using an Extended Kalman Filter (EKF).

Visual Odometry

Mapping was done in 2 ways

  • 3D mapping done using RTAB mapping
  • 2D mapping was done using Hector SLAM and GMapping packages.

Real-Time Appearance-Based mapping

  • As 3D localization and Motion planning was not required 2D grid map was required
  • Dijkstras Algorithm was used as Global Planner
  • Dynamic Window Approach (DWA) planner was used as Local Planner.
  • This came bundled in Nav Stack

Motion Planning Using Nav-Stack

GitHub

Team

  • Karthik Raman
  • Prathamesh RInge
  • Sania Subhedar
  • Sushlok Shah
  • Yagnesh Devada
  • Aayush Fadia
  • Kushagra Srivastva
  • Varad Vaidya
  • Harshad Zade
  • Anshul Paigwar
  • Ajinkya Kamat

Faculty Mentors:
Dr. Shital S. Chiddarwar