Browse Prior Art Database

Method for Robot Path Data Processing and Compensation

IP.com Disclosure Number: IPCOM000248980D
Publication Date: 2017-Jan-25
Document File: 3 page(s) / 77K

Publishing Venue

The IP.com Prior Art Database

Related People

Cheng Li: AUTHOR [+3]

Abstract

Industrial robots work with continuous path in applications such as dispensing, laser cutting. Path programing is not easy, particularly for an irregular path other than a plain circle or rectangle. On the other hand, even with a perfect programing, the actual path may not fit the desired one due to the robot error and motion effect. An external positon measurement device (e.g. laser tracker) can be used for recording a path and identify the path deviation. As filled in previous patent (see the attached patent), ABB closed the idea of using touch panel as position measurement device for generating and compensating 2D robot paths. However, to realize the path generation and compensation, two problems remain open: 1) Measurement device samples a huge path with lots of position data, which can cause path planning error or crash the robot controller. How to process the data with reasonable number of robot targets while still keep the path feature? 2) It is not easy to synchronize the external sampling with robot motion. To identify the path deviation, how to align the recorded path with the programmed robot targets? In this disclosure, we present the method to solve the above mentioned problems. It is worth noting that the method is implemented in touch panel robot programing as verification (see the attached PPT), but it can be generic for robot path processing via any measurement device.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

1

Method for Robot Path Data Processing

1 BACKGROUND Industrial robots work with continuous path in applications such as dispensing, laser cutting. Path programing is not easy, particularly for an irregular path other than a plain circle or rectangle. On the other hand, even with a perfect programing, the actual path may not fit the desired one due to the robot error and motion effect.

An external positon measurement device (e.g. laser tracker) can be used for recording a path and identify the path deviation. As filled in previous patent (see the attached patent), ABB closed the idea of using touch panel as position measurement device for generating and compensating 2D robot paths.

However, to realize the path generation and compensation, two problems remain open:

1) Measurement device samples a huge path with lots of position data, which can cause path planning error or crash the robot controller. How to process the data with reasonable number of robot targets while still keep the path feature?

2) It is not easy to synchronize the external sampling with robot motion. To identify the path deviation, how to align the recorded path with the programmed robot targets?

In this disclosure, we present the method to solve the above mentioned problems. It is worth noting that the method is implemented in touch panel robot programing as verification (see the attached PPT), but it can be generic for robot path processing via any measurement device.

2 THE IDEA

2.1 Curvature Guided Robot Path Data Filtering

With the help of curvature guided data filtering, we down-sample all the position data to a reasonable amount which still can properly fit the original path, thus solve the first problem. Furthermore, we process the data in such a way that robot will only provide linear movement to simplify the robot programming.

The core problem here is whether we should keep or discard each individual point of all the position data , … , . In order to clearly state our method, let’s first define the following two functions:

- Dline( , ) = ‖ − ‖, the Cartesian distance between any adjoining two points;

2

- Darc , = D ( , ) +⋯+ Dline , , the arc length between any two points along the path, which is the sum of all the Cartesian distance of adjoining two points in between.

For a point , we do the following:

Ÿ If it is the first point or the last point , we retain the point;

Ÿ If D ( , ) < 0.999 Darc( , ), which suggests the local curvature is large, then the point will be retained, as illustrated in Figure 1(a); otherwise, we will remove the point since the straight line from to is a good represent...