Skip to content

Latest commit

 

History

History
47 lines (36 loc) · 2.57 KB

File metadata and controls

47 lines (36 loc) · 2.57 KB

Real-Time Visual Pose-Graph-Estimation

Final project for MSAI495: Computer Vision

Author: Sayantani Bhattacharya, Sharwin Patil

We implemented a ROS2 Package that extracts synchronized stereo or monocular raw image frames from an Intel RealSense Camera. Upon initialization, the node configures itself based on runtime parameters including camera mode (stereo/monocular), feature extraction method (ORB/SIFT), and optimization frequency. For stereo operation, it establishes synchronized subscribers for both infrared cameras and their calibration data, while monocular mode uses the color camera stream. For more details visit Portfolio Post

And extended the project to develop a F1 Ghost Car Racing using Visual-Inertial Odometry. A Multi-player game that scores players on how accurate a defined race track is followed. Game Demo

Quickstart for TurtleBot3:

  1. Connect to NUMSR (Automatic no Static IP)

  2.  ssh -oSendEnv=30 msr@teetle #NAME OF TURTLEBOT GOES HERE.
     # Password is robotics!
    
  3. Verify working status: should hear jingle opencr_install_firmware

  4. # Not sure if needed
    source /opt/ros/iron/setup.bash
    source ~/ros_home_ws/install/setup.bash
    
  5. # Run the setup node: 
     source /home/msr/install/setup.bash
     ros2 run numsr_turtlebot numsr_turtlebot
    
  6. # Test for wheel command to Move forward: 
     ros2 topic pub /wheel_cmd nuturtlebot_msgs/msg/WheelCommands "{left_velocity: 100, right_velocity: 100}" 
    

Dependency Installation:

Resource

Hardware:

  • TurtleBot3 [Static IP: 192.168.2.1 | Name: Teelte]
  • RealSense D435I
  • Zed 2i
  • LiPo Battery