Assistance robots are typically mobile robots designed to assist people in shopping malls, airports, healthcare facilities, home environments, and various other environments. Among other things, these robots can help users find their way in unfamiliar environments, for example by guiding them to a specific location or by sharing important information with them.
While the capabilities of assistive robots have improved significantly over the past decade, the systems implemented to date in real-world environments are not yet capable of efficiently tracking or directing people in crowded spaces. In fact, training robots to follow a specific user while navigating a dynamic environment characterized by many randomly moving “obstacles” is far from an easy task.
Researchers at the Berlin Institute of Technology recently introduced a new model based on deep reinforcement learning that could: mobile robots to direct a specific user to a desired location or follow him/her while carrying their belongings, all in a busy environment. This model, introduced in a paper pre-published on arXiv, could help to significantly improve the capabilities of robots in shopping malls, airports and other public places.
“The task of guiding or tracking a human in crowded environments, such as airports or train stations, to carry weight or goods is still an open problem,” wrote Linh Kästner, Bassel Fatloun, Zhengcheng Shen, Daniel Gawrisch and Jens Lambrecht. in their newspaper. † “In these use cases, the robot is required not only to interact with people intelligently, but also to navigate safely among crowds.”
When training their model, the researchers also used semantic information about the states and behavior of human users (e.g., talking, running, etc.). This enables their model to make decisions about how best to help users, by moving next to them at a similar pace and without colliding with other people or obstacles nearby.
“We propose a deep amplification learning-based agent for guiding and tracking tasks in busy environments,” the researchers wrote in their paper. “Therefore, we incorporate semantic information to provide the agent with high-level information, such as people’s social status, safety models, and class types. .”
To test the effectiveness of their model, the researchers conducted a series of tests using arena-rosnav, a two-dimensional (2D) simulation environment for training and assessment. deep learning models† The results of these tests were promising, because in the simulated scenarios, the artificial device could both direct people to specific locations and track them, adapting the speed to that of the user and avoiding nearby obstacles.
“We evaluated our proposed approach against a benchmark approach without semantic information and showed improved navigational safety and robustness,” the researchers wrote in their paper. “In addition, we show that the agent could learn to adapt its behavior to humans, which significantly improves the interaction between humans and robots.”
The deep reinforcement learning The model developed by this team of researchers was found to work well in simulations, so its performance now needs to be validated using physical robots in real environments. In the future, this work could pave the way for creating more efficient robotic assistants for airports, train stations and other crowded public spaces.
Linh Kästner, Bassel Fatloun, Zhengcheng Shen, Daniel Gawrisch, Jens Lambrecht, Human-following and -guiding in busy environments using semantic deep reinforcement learning for mobile service robots. arXiv:2206.05771v1 [cs.RO]† arxiv.org/abs/2206.05771
© 2022 Science X Network
Quote: A model that allows robots to track and guide humans in crowded environments (2022, July 1), retrieved July 1, 2022 from https://techxplore.com/news/2022-06-robots-humans-crowded-environments.html
This document is copyrighted. Other than fair dealing for personal study or research, nothing may be reproduced without written permission. The content is provided for informational purposes only.