When it comes to the evolution of mobile robots, it may be a long time before legged robots can seamlessly interact with the real world, according to new research.
The study was conducted by a team of researchers at Ohio State University recently published in IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2022 describes methods for testing and evaluating the safety of legged robots, machines that, unlike their legged counterparts, rely on active legs for movement. The study found that many current models of legged robots do not always predict responses to real-life situations, meaning it is difficult to predict whether they will fail. – or success – and any given task requires movement.
“Our work reveals that these robotic systems are complex and, more importantly, defy logic,” said Bowen Weng, PhD candidate in electrical engineering and computing at Ohio State. “It means you can’t rely on the robot’s ability to know how to react in certain situations, so the completeness of the experiment becomes even more important.”
AMobile robots are developing to perform various and sophisticated tasks, and many people in the scientific community are realizing that the industry needs a system of global safety testing standards, especially as robots and other intelligences have started slowly flowing into our daily lives. LEgged robots in particular, which are often made of steel and can run as fast as 20 mph, can quickly become a safety hazard. He is expected to work with humans in an unpredictable environment, Weng said.
“Testing is really about risk assessment, our goal is to assess how much risk robotics presents to current employees or customers in the workplace,” he said.
While there are now some safety specifications for deploying legged robots, Weng said there is still no agreement on how to test them in the field.
The study presents the first data-driven, first-of-its-kind safety analysis method for legged robots, Weng said.
“In the future, these robots may have the opportunity to co-exist with humans, and it is likely that many countries around the world will cooperate,” he said. “So having safety standards and testing is critical to the success of this type of product.”
The research, which was inspired by Weng’s work as a car safety researcher and Transportation Research Instituteof partnership and National Highway Traffic Safety Administrationtakes advantage of machine learning-based algorithms to determine how well-designed robots will fare during real-world testing.
Although various factors can be used to demonstrate the safety performance of robots, this study investigated The position of the robot will not fail as you actively navigate the new environment. And because many of the algorithms the team used came from previous robotics experiments, they were able to design multiple scenarios for the simulation to run.
One experiment focused on learning the robot’s ability to adapt to different tasks, such as walking backwards or stepping into position. Elsewhere, researchers tested whether the robot would fall if it was occasionally pushed with enough force to change its direction.
The study showed that while one robot remained upright for 3 out of 10 trials when it was told to accelerate, another could remain upright for over 100 trials when it was driven from his left side, but it failed during 5 out of 10 tests when. the same force is applied to its right side.
Ultimately, the researchers’ method could help support the commercialization of legged robots and help establish safety standards for robots with different structures and properties, although Weng said it would take shortly before it can be applied.
“We believe that this data-driven approach will help to create an impartial, more effective way to conduct research of robots in the context of research areas,” he said. “What we’re working on is not immediate, but for researchers down the line.”
The authors are Guillermo Castillo and Ayonga Hereid of Ohio State, and Wei Zhang of the school. Southern University of Science and Technology in Shenzhen, China. This work was supported by the National Science Foundation and the National Natural Science Foundation of China.
