kc_vision — Autonomous ground robot with SLAM, Nav2 + GPS boustrophedon coverage planner (ROS 2 Jazzy + Gazebo Harmonic)

Projects
, , , ,
1

Hi ROS community! I’m sharing kc_vision, a fully open-sourced autonomous ground robot project I built during my M.Tech in Robotics & AI. The complete simulation stack runs out of the box on Ubuntu 24.04 — no hardware needed to try it.

What it does

  • Builds its own maps (SLAM Toolbox or Google Cartographer)
  • Navigates autonomously to clicked goals in RViz (Nav2 + obstacle avoidance)
  • Follows GPS waypoints outdoors without a pre-built map
  • Sweeps entire outdoor areas with a boustrophedon coverage planner — mark 4 GPS corners, it handles the rest

Tech stack

  • ROS 2 Jazzy
  • Gazebo Harmonic
  • Nav2
  • SLAM Toolbox / cartographer_ros
  • robot_localization (EKF sensor fusion)
  • ros2_control (differential drive)
  • micro-ROS (planned for hardware phase)

The GPS coverage planner (the interesting part)

Drive to 4 corners of an area, log the GPS waypoints with gps_waypoint_logger.py, then run gps_coverage_planner.py. It computes parallel sweep rows (0.2 m spacing, configurable) and autonomously drives the entire boustrophedon pattern. No map needed — EKF sensor fusion via robot_localization keeps the position accurate outdoors.

Hardware (parallel build in progress)

Raspberry Pi 5 · STM32 via micro-ROS · YDlidar · BNO086 · NEO-M8N GPS
Mechanical design done .

GitHub: git@github.com:bingikiranchand932-hub/kc_vision_ws.git

Happy to answer any questions about the GPS coverage planner implementation or the EKF sensor fusion tuning!

cc: @smac2 @SteveMacenski @JWhitleyWork @joshnewans

ROS News for the Week of July 6th, 2026
2

Hi, you’ve shared the private link to your GitHub. Please, share the HTTPS URL.

Also, when you want to showcase anything in robotics, it’s best to add at least a picture, or - better - a video.

Last, it’s not okay to mention people with @ unless you’re 100% sure they’re interested.

2 Likes