Detecting negative obstacle using Kinect sensor

Muhammad Fahmi Abdul Ghani, Khairul Salleh Mohamed Sahari

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

A robot must have a good understanding of the environment for autonomous navigation. Mobile robot using fixed laser range scanner can only detect obstacle on a plane level. This may cause important obstacles not to be appropriately detected. This will cause the map generated to be inaccurate and collision may actually occur during autonomous navigation. Microsoft Kinect is known to provide a low-cost 3-D data which can be used for mobile robot navigation. Many researchers focused on obstacles above ground level, and not negative obstacles such as holes or stairs. This article proposes the usage of Kinect sensor to detect negative obstacles and converts it into laser scan data. Positive obstacle is defined as the obstacle above the floor surface and negative obstacle is defined as the obstacle below the floor surface. Projection method is used to convert positive obstacle data from Kinect sensor to laser scan data. For negative obstacles detection, farthest point method and virtual floor projection method are used. The laser scan data from positive and negative obstacles are then combined to get an improved laser scan data, which includes all obstacles that are important for a robot to see. The negative obstacle detection methods are tested in simulated indoor environment and also experimental in a real environment. The simulation and experimental results have demonstrated the effectiveness of our proposed method to detect and map negative obstacles.

Original languageEnglish
JournalInternational Journal of Advanced Robotic Systems
Volume14
Issue number3
DOIs
Publication statusPublished - 06 Jun 2017

All Science Journal Classification (ASJC) codes

  • Software
  • Computer Science Applications
  • Artificial Intelligence

Fingerprint Dive into the research topics of 'Detecting negative obstacle using Kinect sensor'. Together they form a unique fingerprint.

  • Cite this