Visual Navigation System for Triceratops Robot
Developing a visual navigation system for autonomous navigation of a Triceratops robot in indoor environments.
Overview
This project demonstrates how the Triceratops robot uses a visual navigation system to navigate indoor environments, with Visual SLAM and AprilTag localization for accurate pose estimation and map building.
Key Results
My Contribution
My Contribution — Stanley Chueh 闕楷宸
- Designed and developed the visual navigation system (Visual SLAM + AprilTag-assisted localization)
- Integrated the visual navigation system with the physical Triceratops robot for autonomous indoor navigation
Collaborators
City Science Lab robotic team built the Triceratops robot and its low-level gait control — not developed by me
External / Open-Source Methods Used
NVIDIA Isaac ROS Visual SLAM · AprilTag
Robot Setup
We use the Triceratops robot platform equipped with a RGB-D camera for visual navigation.
The robot is built by City Science Lab@Taipei Tech Robotic team, and the low-level control(gait control) is not developed by me, but I developed the visual navigation system and integrated it with the robot for autonomous navigation.
Triceratops robot platform setup
System Architecture
The visual navigation system consists of several key components, including Visual SLAM for map building and localization, and AprilTag markers for accurate pose estimation and map switching.
Visual navigation system architecture diagram
Visual SLAM
The visual slam we are using is from NVIDIA Isaac ROS VSLAM, and I take visual odometry into account when integrating the visual navigation system with the triceratops robot for autonomous navigation in indoor environments.
Visual SLAM demonstration for map building and localization
AprilTag-assisted Localization
As the pure visual navigation system can be noisy and not robust in complex environments, we utilize AprilTag markers for accurate localization during indoor navigation.
AprilTag marker detection
AprilTag-assisted localization demo