Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Method and System for Estimating Accurate Real-Time Location of an Autonomous Robot

IP.com Disclosure Number: IPCOM000247932D
Publication Date: 2016-Oct-11
Document File: 6 page(s) / 238K

Publishing Venue

The IP.com Prior Art Database

Abstract

A method and system is disclosed for estimating accurate real-time location of an autonomous robot.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 01 of 6

Method and System for Estimating Accurate Real-Time Location of an Autonomous Robot

In many machine vision applications, fast and accurate determination of position and orientation is frequently needed. Current technology enables tape libraries to utilize autonomous mini-robots shuttling around to pick and place spools to and from a tape reading unit.

Disclosed is a method and system for estimating accurate real-time location of an autonomous robot. The method and system enables quick and accurate determination of the position and orientation of the autonomous robot using a camera and a reference pattern.

The method and system uniquely determines the position of the autonomous robot on a planar 2-Dimensional (2-D) surface by estimating three pose parameters, x, y, and q. In a tape library, posts are distributed throughout the surface, typically in regular pattern, to hold tapes. The posts can be positioned very accurately using mechanical methods. A

simple pattern can be put on each post that allows the autonomous robot to estimate the position and orientation from a single camera view. The method and system uses a colored 90 degree cross pattern. The cross pattern can also be very accurately positioned on the post. The method and system uses an approach that consists of segmenting the image to a binary image with the landmark only. Thereafter, a coarse estimate of the position and orientation of the landmark is found and the coarse estimate is refined to an accurate estimate.

In one implementation, the method and system uses a colored 90 degree cross as illustrated in Fig. 1. The method and system uses a colored cross enabling a fast segmentation. The method and system uses a simple threshold test of comparing if "((red > green) && red - (green+blue)/2 > 20)". The binary segmented image is then component labeled and a rectangular bounding box of an object with the largest area or perimeter is used as a region of interest (ROI). The borders of the ROI are examined to obtain a coarse estimate of the cross endpoints. The pairwise distance of the cross endpoints is computed and the largest distance is used to define a major axis. A minor axis is found by searching an

endpoint list for the longest perpendicular line to the major axis. In the Fig. 1, the ROI is displayed as a red rectangle. The major axis is displayed as a red line and the minor axis as a green line. The display of ROI and the axes result in a coarse estimate of the position and orientation, as can be seen in the Fig. 1. But the coarse estimate can be off by several pixels. If the cross is fully visible, the method and system robustly finds the major and minor axis.

1


Page 02 of 6

Figure 1

Under occlusion the major and minor axes may be swapped as illustrated in the Fig. 2. The swapping can be resolved by

2


Page 03 of 6

com...