Projects(Opensourced) | Swastik Mahapatra

work

3D Gesture Recognition For Robotic Manipulator Control

This 3D Gesture Recognition package is an ROS (Robot Operating System) solution designed for real-time Hand Gesture Reognition. Leveraging the power of deep learning, this package processes depth images captured by ADI's Time of Flight (ToF) sensor to accurately recognize and publish one of 18 distinct gestures. At the heart of the recognition module is a Convolutional Neural Network (CNN), which ensures the system to operate seamlessly at 5 frames per second (FPS) on the EVAL-ADTF3175D-NXZ platform.

3D Depth Stitching to enhance Time-Of-Flight sensors capabilities

This is a ROS (Robot Operating System) package for stitching depth images from multiple Time-Of-Flight sensors like ADI’s ADTF3175D ToF sensor. This node subscribes to captured Depth and IR images from multiple ADI 3DToF ADTF31xx nodes, stitches them to create a single expanded panoramic field of view and publishes the stitched Depth, IR and PointCloud as ROS topics.

Safety Bubble Detector for Industrial AGV

This is a ROS (Robot Operating System) package for the Safety Bubble Detection application. The Safety Bubble Detectors are the basic building block of any AGV/AMR. The safety zone is a virtual area around an AGV/AMR. The Safety Bubble Detectors are used to detect the presence of any object inside this zone and prevent the AGV/AMR from colliding on to the object.

Fast Floor segmentation for Industrial AGV

This is a ROS (Robot Operating System) package for the Floor Detection application. The term "Floor Detection" refers to determining where the floor is in the given image. It is an image segmentation problem in which a given image is divided into floor and non-floor pixels. Floor Detection is an essential component of real-world applications such as Robot Navigation, Autonomous Driving, Augmented reality (AR) applications, and 'Obstacles Detection and Avoidance' for robots and people with inadequate vision.

ROS Node for Time of Flight sensors

The ADI 3DToF ADTF31xx is a ROS (Robot Operating System) package for working with ADI’s ADTF3175D ToF sensor. This node captures the Depth and IR frames from the sensor and publishes them as ROS topics. The node uses ADI ToF SDK APIs to capture frames from the sensor. The node publishes Depth and IR Images at 512x512 (16 bits per image) resolution and Camera Info @ 30FPS.

thesis

Leveraging Deep-Learning for Threat Estimation

This project aims to promote AI for Good by exploring the Global Terrorism Database created by the University of Maryland, curating it, and trying to map untraceable patterns between different terrorism incidents using Artificial Neural Networks. We used this setup to find out the vulnerability of various Indian states to possible criminal activities at different intervals of time. We developed the complete ANN training and inference set-up from scratch to better Learn Deep-Learning Technology.