RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2

LETI: Robot artist

Product
Developers: LETI St. Petersburg State Electrotechnical University
Date of the premiere of the system: October 2022
Branches: Entertainment, leisure, sports

Content

2024: Obtaining new algorithms: they allow the robot to write in calligraphic "handwriting"

Russian scientists from the St. Petersburg State Electrotechnical University "LETI" have developed a robot artist capable of reproducing the technique of applying strokes of famous painters when creating copies of paintings. New algorithms developed by scientists of St. Petersburg State Technical University "LETI" will also allow a robotic painter to increase the accuracy of imitating the style of famous artists. This became known on October 1, 2024.

The study was funded by the Russian Science Foundation, and its results are published in the journal Robotics. The robot is a manipulator moving in three-dimensional space and capable of holding a conventional paint brush. The device is fixed on a table where the canvas is located.

source = LETI
Robot artist

The project manager and senior researcher of the Youth Research Institute of St. Petersburg State Technical University "LETI" Artur Karimov announced the completion of an important stage to improve the robotic painter.

File:Aquote1.png
In the future, we will work on integrating the latest version of the robotic arm and the previously created paint mixer in order to "add colors" to its new abilities, the scientist noted.
File:Aquote2.png

The development of the robot artist began in 2016. The first prototype could only draw black and white paintings using the grisaille technique. Then the researchers created a special acrylic paint mixer, which allowed the robot to receive almost any color and shade for painting.

A new mathematical model developed by the scientists allowed the robot to apply strokes of varying length, width and shape to the canvas. In addition, the device has learned to vary the thickness of the strokes depending on the force of pressing the brush on the canvas. These improvements greatly expanded the functionality of the mechanical painter, allowing him to more accurately reproduce the style of specific artists.

source = LETI
Example of robot operation

Special software was created for the robot. The algorithm breaks the uploaded image into separate strokes and converts this data into machine code, which is then used to control the robotic arm.[1]

2022: Robot Artist Presentation

At the end of October 2022, LETI presented a hardware robot artist at St. Petersburg State Electrotechnical University, which, according to the developers, can be used to restore and copy paintings of any complexity.

The approximate cost of the car is more than 200 thousand rubles, told TASS developer and associate professor of the Department of Computer Aided Design Systems of St. Petersburg State Technical University "LETI" Arthur Karimov.

Hardware robot artist presented in St. Petersburg

He also noted that the robot's technology is based on applying swabs using a given algorithm using a gradient. It was a difficult task to get to work with paints, but it was solved. Now specialists will work to ensure that the quality of transmission of paints and smears improves, Karimov said.

It is known that the robot has predecessors that were developed by other scientists. Previously, experts only programmed the robot to mix colors on the palette, so it was difficult to convey clean colors.

File:Aquote1.png
Going to work with paints was a difficult task, but we decided it, in the future we will work to improve the quality of transmission of paints and strokes, I think in the horizon of five years we will be able to copy some work of art of some given era. Now we work only with acrylic paints, in the future we plan to switch to oil, - said the associate professor of the Department of Computer Aided Design Systems of St. Petersburg State Technical University "LETI."
File:Aquote2.png

According to him, in the future it will also be possible to set any playback accuracy parameters. By October 2022, the error in copying is approximately 7%, he said. Karimov added that in order to achieve the maximum effect, it will be necessary to use authentic paints, for this it is necessary to produce machine learning for each type separately.[2]

Notes