Researchers are developing a new AI form that can be modified to perform transferable tasks TOU

Researchers are developing a new AI form that can be modified to perform transferable tasks


Researchers are developing a new AI form that can be modified to perform transferable tasks

Robot Diego is ready to stack the cubes. Credit: Maximilian Diehl

Can robots modify their own work patterns to solve complex tasks? Researchers at the Salmers University of Technology in Sweden have developed a new form of AI that can observe human behavior and perform its functions in an adaptive environment. The hope is that robots that can be flexible in this way will be able to interact with a greater number of humans.

“Robots working in human environments must be adapted so that humans are unique and we can all solve the same task differently. Therefore, an important part of robot development is to teach robots how to interact with humans.” Maximilian Deal, a graduate student and key researcher behind the project.

When humans perform a simple task such as setting a table, they can approach the challenge in a variety of ways depending on the circumstances. If a chair unexpectedly stands in the way, we can choose to move it or walk around it. We alternate between using our right and left hands, pausing, and performing many unplanned actions.

But robots do not work the same way. They need precise programming and methods to achieve the goal. This approach makes them more efficient in environments that follow the same pattern as factory processing taxes. But in order to successfully communicate with those in areas such as health or customer-facing roles, robots need to create more flexible work ways.

“In the future, we expect robots to do some basic household chores, such as setting the table, cleaning kitchen utensils, or arranging groceries,” said Karen Ramirez-Amaro, an assistant professor in the Department of Electrical Engineering.

Researchers at the University of Salmers wanted to explore whether a robot could be taught human-like approaches to solving tasks – to create a “descriptive AI” that could extract what was common instead of specific information during a demonstration so that it could be flexible and planned. The adaptive path towards a long-term goal. Explainable AI (XAI) is a term used to describe a type of artificial intelligence that allows humans to understand how a particular end or end was achieved.

This video is part of the “Automated Generation of Robotic Planning Domain from Observations” release by Maximilian Diehl, Chris Paxton and Karin Ramirez-Amaro. Credit: Rational Lab

Teaching a robot to stack objects under changing conditions

Twelve times in the VR environment, researchers were asked to do the same task of stacking small cubes. Each time the task was done in a different way, and the movements made by humans were monitored by laser sensors.

“When we humans have a task, we break it down into smaller chains, and every action we take is aimed at accomplishing the intermediate goal. Instead of teaching the robot to accurately follow human behavior, we focus on figuring out what the goals are and all the actions that those involved have done. Seeing, ”says Carinne Ramirez-Amaro.

The unique approach of the researchers was to focus AI on extracting the purpose of the sub-targets and creating libraries with different functions for each. Later, AI developed a scheduling tool that could be used by the TIAGo robot – a mobile service robot designed to work in indoor environments. With the help of the tool, the robot was able to automatically create a plan for the task of stacking the cubes on top of each other, even if the surrounding conditions changed.

In summary: The task of stacking the cubes was given to the robot, who then selected a combination of several possible actions to create an array that would lead to the completion of the task, depending on the circumstances, which changed slightly for each attempt. . The results were very successful.

“Through our AI, the robot produced projects with a 92% success rate after one human demonstration. When information from all twelve demonstrations was used, the success rate reached 100%,” says Maximilian Diehl.

The work was presented at the robot conference IROS 2021, one of the most prestigious conferences in the world related to robotics. In the next phase of the project, researchers will explain how robots can interact with humans and what went wrong, why, if they fail in a task.

Occupation and health

The long-term goal is to use robots in tasks such as tightening bolts / nuts on truck wheels to assist technicians in tasks that can cause long-term health problems. In health care, there may be tasks such as carrying and storing medicine or food.

“We want to facilitate the work of health professionals so that they can focus on tasks that require more attention,” says Karin-Ramirez Amaro.

“It may take many more years for us to find real autonomous and multi-purpose robots, especially as individual challenges such as computer vision, control and secure communication with humans remain to be solved. However, we hope that our approach will contribute to accelerating the learning process of robots, incorporating all these new features. It also allows the robot to use them in situations, “said Maximilian Deal.

                                    <div class="article-main__explore my-4 d-print-none">

                                        A new framework that facilitates the learning of hue in robotics
                                    <hr class="mb-4"/>
                                                                                                                        <div class="article-main__more p-4">
                                                                                            <strong>More info:</strong>
                                            read more <a href="" target="_blank" rel="noopener"></a>


                                                                                <div class="d-inline-block text-medium my-4">

                                            Presented by Salmers University of Technology
                                                                                                    <a class="icon_open" href="" target="_blank" rel="nofollow noopener">
                                                        <use href="" x="0" y="0"/>


                                    <!-- print only -->
                                    <div class="d-none d-print-block">
                                             <strong>Quote</strong>: Researchers develop new AI form that can work in changing environments (April 14, 2022)

                                        This document is subject to copyright.  No part may be reproduced without written permission, except for any reasonable manipulation for the purpose of personal study or research.  Content is provided for informational purposes only.



if you want to read this article from the original credit source of the article then you can read from here


Leave a Reply