When it comes to the development of mobile robots, it may be a long time before legged robots can safely interact in the real world, according to a new study.
Led by a team of researchers at The Ohio State University, the study published recently in the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2022 describes a framework for testing and characterizing the safety of legged robots, machines that, unlike their wheeled counterparts, rely on mechanical limbs for locomotion. The study found that many current legged robot models do not always act predictably in response to real-world situations, meaning it is difficult to predict whether they will fail. – or succeed – at any task that requires movement.
“Our work reveals that these robotic systems are complex and, more importantly, counter-intuitive,” said. Bowen Weng, PhD student in electrical and computer engineering 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 testing becomes even more important.”
AAs mobile robots evolve to perform more diverse and complex tasks, many in the scientific community also realize that the industry needs a set of universal safety testing regulations, especially as robots and other artificial intelligence have gradually started to flow into our daily lives. Legg-filled robots in particular, which are often made of metal and can run as fast as 20 mph, could quickly become safety hazards when expected to work alongside people in real and often unpredictable environments, Weng said.
“Testing is really about assessing risk, and our goal is to investigate how much risk robotics currently poses to users or customers during a working condition,” he said.
Although there are currently some safety specifications for the deployment of legged robots, Weng noted that there is still no common agreement on how to test them in the field.
This study develops the first data-driven, scenario-based safety testing framework of its kind for legged robots, Weng said.
“In the future, these robots may have the opportunity to live with humans side by side, and most likely will be produced collaboratively by several international parties,” he said. “So having safety and testing rules is extremely important to the success of this type of product.”
The research, which was partly inspired by Weng’s work as a vehicle safety researcher at the Transportation Research Centerwho partners with the National Highway Traffic Safety Administrationleverages sample-based machine learning algorithms to discern how simulated robots would fail during real-world testing.
Although various factors can be used to characterize the overall safety performance of a robot, this study analyzed a set of conditions under which the robot would not fall while actively navigating a 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 simulations to run.
One test focused on studying the robot’s ability to move while performing tasks at different gaits, such as walking backwards or stepping in place. In another, researchers tested whether the robot would fall if it was periodically pushed with enough force to change its direction.
The study showed that while one robot failed to remain upright in 3 out of 10 trials when asked to slightly accelerate its gait, another was able to remain upright for 100 trials when pushed from its left side, but fell over in 5 out of 10 trials when the same force was applied to its right side.
Ultimately, the researchers’ framework could help witness the commercial deployment of legged robots and help establish a safety benchmark for robots created with different structures and properties, although Weng noted that it will take some time before it can be implemented.
“We believe this data approach will help create an unbiased, more efficient way to make observations of robots in the conditions of a test environment,” he said. “What we are working towards is not immediate, but for researchers down the line.”
Co-authors were Guillermo Castillo and Ayonga Hereid of Ohio State, and Wei Zhang of the 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.
