This repository contains a series of small tests with the universal robots UR-5 robot. The tests might not be immediately runnable on your computer without a bit of effort. On top of that the code isn't really cleaned up. It's more or less a personal history while trying to figure how to use the robot and what to do with it.
The code is written either in Python or C++, utilizing the following libraries:
- python-urx to drive the robot from python https://github.com/jkur/python-urx.git
- ur_rtde to drive the robot from python/c++ https://gitlab.com/sdurobotics/ur_rtde
- cmake to build c++ projects https://cmake.org/
- glm for vector math in c++ https://glm.g-truc.net
- numpy and scipy for vector maths in python https://numpy.org/ https://www.scipy.org/
- opencv for image recognition https://opencv.org/
- gphoto2 for camera tethering http://www.gphoto.org/
I use the fork from jkur, install it using:
pip3 install git+https://github.com/jkur/python-urx.git
import urx
a = 0.1
v = 0.1
rob = urx.Robot("192.168.0.100");
rob.movel((+0.1,-0.5,0.05, 0, 0, 0), a, v)
rob.movel((+0.1,-0.5,0.00, 0, 0, 0), a, v)
rob.movel((-0.1,-0.5,0.00, 0, 0, 0), a, v)
rob.movel((-0.1,-0.5,0.05, 0, 0, 0), a, v)
rob.close()
ur_rtde is a library for both python and C++.
The library itself is great, but unfortunately pybind11 dependency was acting strange on my computer. For me this was the setup process:
pip3 install --user pybind11
cd /usr/local/include/python3.7m
ln -s /Users/hansi/Library/Python/3.7/include/python3.7m/pybind11 pybind11
You can figure out where python3 thinks pybind11 should be available by running
run python3, then:
import pybind11
pybind11.get_include()
pybind11.get_include(True)
This lists two paths, in my case /usr/local/include/python3.7m/pybind11 and /Users/hansi/Library/Python/3.7/include/python3.7m/pybind11. Make sure the header files are present in both folders!
Once this is settled you can run
pip3 install ur_rtde
Uses python and python-urx and does simple motions only. There are three examples inside:
- test_1: Moves the robot to different points
- test_2: Draws a grid
- test_3: Draws the xy-representation of a stereo audio file
- test_2_1: Moves the tool point to different orientations using python-urx
- test_2_2: Moves to different points in space using the python binds of ur_rtde
Uses C++, cmake and ur_rtde to make smooth robot motions.
Uses C++, cmake and ur_rtde to draw with a swerving pen motion attempting to resemble the motion of a human hand. The basis for the drawings are again stereo audio files interpreted as xy-curves.
Moves the robot up, then lets the arm fall freely until it hits the ground. This could be used to quickly measure the length of a pen the gripper is holding.
Testing to see if it's possible to "poke" a human using the robot arm. I consider this experiment an utter failure.
Puts the robot into freedrive. It then records data points until the human stops moving the robot. After that the motion is replayed.
An attempt to make robot move in the way a snake would. I consider this another undisputable failure.
Attached a standard SLR camera to the robot arm using a 3d printed tool adapter. gphoto2 is used for tethering.
Here gummi bears are picked and placed. The gummi bears are detected using opencv, Charuco markers are used to align the opencv camera view to the robot coordinate system.
The C++ part handles image recognition and coordinate conversion. Python access the coordinates using a rest interface. Very much in progress.
Very much in progress. This is a continuation of test_8_camera, but for the laowa macro probe lense and ... much more complicated of course.