M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion Simulator of a SCARA Robot with LABVIEW...

Preview:

Citation preview

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Simulator of a SCARA Robot with LABVIEW

Ing. Luca Baglivo, Ing. Alberto Fornaser

Department of Structural Mechanical Engineering, University of Trento

Email: a.fornaser@ing.unitn.it

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

INTRO

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Simulink Diagram

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Simulink Diagram

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Time

Trajectory

Speed of theEnd Effector

Using the q it is possible to retrive the pose of the robot and the carthesian coordinate of the End Effector.

Jacobian evaluation

Compensate the error of the position due to numerical integration

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

LabView Diagram

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

LabView Diagram: GUI

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Main Logic Blocks

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Main Logic Blocks

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Simulator

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Software Main Structure

The software is thought in order to pilot a robot.It is a good approach organize the software it in the following way: - First Task -> Move the robot to the HOME position (safe pose); - Second Task -> Move the robot, our simulatoin; - Third Task -> Go back to HOME.

This logic can be achieved through a Sequence structure.

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Trajectory

-Time dependent

Input:Center xCenter yRadiusT

Output:x(t) y(t)

)/cos()( tc ktrxtx

)/sin()( tc ktryty

Name: Trajectory.vi

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Trajectory

The robot must be settled avoiding a singular initial pose. In such way it is possible to achieve and follow a feasible path without problems

Initial Conditions:Q1 -> -pi/2Q2 -> pi/2Q3 -> 0

A1

A2

Geometry of the robot:A1 -> 300mmA2 -> 350mmA3 -> 0mm

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

ROBOT VISUALIZATION

This Vi helps the user to understand the pose of the robot and monitor the evolution of the motion.We need to have the full structure as output, this will simplify the programming interface a lot.

Arm_visualizer.vi

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Jacobian and Pseudoinversion

This Vi must evaluate the Jacobian Matrix. The input are the angles of the joints.

The output should be a matrix, and this should be inverted in a second Vi.

NB: J is related to the dimension of the joints, so it is usefull to create a Vi that output these informations.

1)'(' JJJJ inv

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Derivator

This Vi must evaluate the velocity of the End Effectot

Important: during the first loop we cannot evaluate the speed, so we need to pass directly a value.

This VI must have 2 additional imput, 2 single value ( double ). These inputs will be used later to introduce a numeric error compensation.

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Integrator

This VI is used after the multiplication between the presudoinverse Jacobian and the End Effector Velocity Vector. By integrating the output of that multiplication the qs (angles of the joijts) of the robot can be evaluated.

Using the Output of this Vi it is possible to connect the Robot Visualization Vi, and watch the motion of the robot directly during the processing loop.

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Images: Some Hints

A Good methodology in LabView is to start building your Vi referring to an example, copying part of the code and then start adding your own «code».

So, how can we open an Image in Labview?

Go to:

…Program Files (x86)\National Instruments\LabVIEW 2010\examples\Vision\1. Getting Started

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

How Initialize and Load an Image

We should works with 2 images at the same time:-One is the source image-One is the destination Image

This is due to the library related to the image processing. LabView does not allow the user to overwrite an image from the same function.

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Camera View Simulation

Vi that simulates the camera mounted on the End Effector

Input:- Image of the ground- Coordinates of the End Effector

Vi that performs the chek of the image, the task is to identify the center of a dot seen by the camera.Use Vision Assistant!

Vi that returns the coordiante of the identified dot in the world coordiante system

GIVEN

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Camera Simulation

The simulation is based on the identification of a region of 640x480 pixels centered and oriented according to the actual pose of the End Effector.

NB: Camera and Robot Reference Coordinate Systems are different!

World RCS

Camera RCS

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Camera Simulation

One of the limits of the graphic library in LabView is that you can not crop region of image using any orientation.

To overcome this issue is possible to shift and rotate the image.

Region to Crop

ROI Centered to the origin 0,0 with dimensions 800x800 pixels

Why 800x800?

In this region i can have all the possible 640X480 sub regions with any generic orientation!

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Camera Simulation

Region to Crop

Region to Crop

We shift the image, not the reference system!

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Camera Simulation

Now we can apply the rotation to the image.

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Camera Simulation

We can now Crop the image to the 640x480 region.This is the resulting image that simulates the view of the camera!

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Camera Simulation

NB: The rotation of the image is done around the center of the image itself.

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Vision Assistant

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Vision Assistant

Operation Order

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

From Camera to World RCS

Now we have the estimation of the position of the dot expressed in the camera reference coodinate system, but the robot usually works inside its own space (called also World).

We need a Vi that performs the transformation between the camera rcs and the world rcs.

Dot identified inside the image Field of view of the camera inside the world reference system

Coordinates of the dot expressed in the world system

Camera Reference System

M. De Cecco - Lucidi del corso di Robotica e Sensor Fusion

Recommended