/ros_object_det_motor_control

Primary LanguagePython

OBJECT DETECTION BASED MOTOR CONTROLLER.

This is a ROS node for controlling a motor based on object detection. It works as:

  • Reads each RTSP link from MongoDB.
  • Creates respective Image publisher for each RTSP link.
  • Object detection on each image topic identified.
  • Publish motor status based on object detection
  • Controlling motor using Twist configuration.

Requirements

  • ROS Noetic
  • Python 3.9
  • pip3

Installation:

  1. Clone the repository into your ROS workspace.

  2. Install dependencies.

     pip3 install -r scripts/requirements.txt

  3. Install using docker:

    1. build docker

      docker build -t my_dock .

    2. Start roscore in your local machine:

      roscore

    3. start the docker:

      docker run -it  --network="host" --env="ROS_IP"=$ROS_IP  --env="ROS_MASTER_URI"=$ROS_MASTER_URI my_dock roslaunch motor_control main.launch

    4. To Visualise these videos. Start rqt image viewer in local machine:

      rqt_image_view

Usage:

  • db_utils.py : Use to manage Database.
  • camera_publisher.py : reads RTSP/video file path from Mongodb database and publishes streams.
  • object_detection.py : Subscribes to above published streams, performes object detection and publihes it's result to /motor_status topic.
  • motor_control.py : Gets the decision from object detection and sets motor's parameters accordingly.

Note:

  • Instead of real RTSP links, video file path in MongoDB database are been used.
  • Output should be like when man walks in frame - Motor status is OFF else ON (when no man in image).
  • For more technical details please look at report.pdf

Architectring.

Architect