Quadruped Inspection Robot System

Featured

ROS-based inspection system for electrical substations using Deep Robotics X30, thermal imaging, and LiDAR-Inertial Odometry.

  • ROS2
  • Python
  • C++
  • YOLOv8
  • FLIR
  • Deep Robotics X30
  • LIO

Industrial-grade autonomous inspection system built on the Deep Robotics X30 quadruped platform, deployed for electrical substation inspection at PLN, Pertamina oil & gas facilities, and PTPN plantations.

🎯 Mission

Replace manual inspection walks in hazardous electrical substations with an autonomous robot that can:

  • Detect overheating components via thermal imaging
  • Navigate complex outdoor terrain with precise localization
  • Identify PPE compliance for personnel entering hazardous zones
  • Transmit real-time telemetry to operators via web dashboard

System Components

Hardware Platform

  • Base Robot β€” Deep Robotics X30 quadruped (also Lite3 for indoor tasks)
  • Thermal Camera β€” FLIR for component temperature monitoring
  • LiDAR β€” 3D point cloud for navigation and mapping
  • RTK GPS β€” Centimeter-level outdoor localization

Software Architecture

text
Robot Hardware (X30)
      ↓
  ROS2 Middleware
      ↓
 β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”  β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”  β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
 β”‚ LIO Nav    β”‚  β”‚ PPE Detectionβ”‚  β”‚ Thermal Detectβ”‚
 β”‚ (GPS+LiDAR)β”‚  β”‚ (YOLO v8)   β”‚  β”‚ (FLIR + CV)   β”‚
 β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜  β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜  β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
      ↓
  Mission Controller
      ↓
  Web Dashboard (Next.js + WebSocket)

Key Technical Work

LIO Navigation (LiDAR-Inertial Odometry)

Resolved significant challenges with LIO in open outdoor environments (GPS-available) by fusing RTK GPS corrections into the LIO state estimate:

  • Drift correction during long traversals
  • Seamless indoor/outdoor transition
  • Reliable performance in high-electromagnetic-interference environments (substations)

PPE Detection Application

Implemented YOLOv8-based PPE detection to verify personnel are properly equipped before entering hazardous zones. Detects: hard hats, safety vests, safety boots.

Custom GUI Interface

Built operator-facing GUI for:

  • Live video feeds (thermal + RGB)
  • Robot state visualization
  • Waypoint mission programming
  • Anomaly logging and reporting

πŸ›  Tech Stack

  • ROS2 β€” robot middleware
  • Python + C++ β€” navigation and perception nodes
  • YOLOv8 β€” real-time object detection
  • FLIR SDK β€” thermal camera integration
  • Next.js β€” web monitoring dashboard
  • WebSocket β€” real-time telemetry streaming

Publications

This work contributed to the paper:

"Development of a Quadruped-Legged Robot for Position Estimation of Overheat Components in Electrical Substations Using a Thermal Camera"

Published as part of ongoing research at PT. Ezra Robotics Teknologi.

More projects

  • Web

    Interactive OS-themed developer portfolio with draggable windows, Spotify integration, i18n, and dark mode.

    • Next.js
    • TypeScript
    • Tailwind CSS
    • Zustand
    • +2 more
  • ROS-based autonomous navigation system for Autonomous Surface Vehicle with multi-sensor fusion, winning Grand Champion at IRC 2023.

    • ROS Noetic
    • Python
    • C++
    • OpenCV
    • +3 more