RT/ A new twist on artificial ‘muscles’ for safer, softer robots
Robotics & AI biweekly vol.97, 1st July — 19th July
TL;DR
- Engineers have developed a new soft, flexible device that makes robots move by expanding and contracting — just like a human muscle. To demonstrate their new device, called an actuator, the researchers used it to create a cylindrical, worm-like soft robot and an artificial bicep. In experiments, the cylindrical soft robot navigated the tight, hairpin curves of a narrow pipe-like environment, and the bicep was able to lift a 500-gram weight 5,000 times in a row without failing.
- Artificial intelligence chatbots have frequently shown signs of an “empathy gap” that puts young users at risk of distress or harm, raising the urgent need for “child-safe AI,” according to a study.
- Computer scientists have invented a camera mechanism that improves how robots see and react to the world around them. Inspired by how the human eye works, their innovative camera system mimics the tiny involuntary movements used by the eye to maintain clear and stable vision over time.
- A technique can plan a trajectory for a robot using only language-based inputs. While it can’t outperform vision-based approaches, it could be useful in settings that lack visual data to use for training.
- Meet CARMEN, short for Cognitively Assistive Robot for Motivation and Neurorehabilitation — a small, tabletop robot designed to help people with mild cognitive impairment (MCI) learn skills to improve memory, attention, and executive functioning at home.
- And more!
Robotics market
The global market for robots is expected to grow at a compound annual growth rate (CAGR) of around 26 percent to reach just under 210 billion U.S. dollars by 2025.
Size of the global market for industrial and non-industrial robots between 2018 and 2025 (in billion U.S. dollars):
Latest News & Research
A Flexible, Architected Soft Robotic Actuator for Motorized Extensional Motion
by Taekyoung Kim, Pranav Kaarthik, Ryan L. Truby in Advanced Intelligent Systems
Northwestern University engineers have developed a new soft, flexible device that makes robots move by expanding and contracting — just like a human muscle.
To demonstrate their new device, called an actuator, the researchers used it to create a cylindrical, worm-like soft robot and an artificial bicep. In experiments, the cylindrical soft robot navigated the tight, hairpin curves of a narrow pipe-like environment, and the bicep was able to lift a 500-gram weight 5,000 times in a row without failing.
Because the researchers 3D-printed the body of the soft actuator using a common rubber, the resulting robots cost about $3 in materials, excluding the small motor that drives the actuator’s shape change. That sharply contrasts typical stiff, rigid actuators used in robotics, which often cost hundreds to thousands of dollars. The new actuator could be used to develop inexpensive, soft, flexible robots, which are safer and more practical for real-world applications, researchers said.
“Roboticists have been motivated by a long-standing goal to make robots safer,” said Northwestern’s Ryan Truby, who led the study. “If a soft robot hit a person, it would not hurt nearly as much as getting hit with a rigid, hard robot. Our actuator could be used in robots that are more practical for human-centric environments. And, because they are inexpensive, we potentially could use more of them in ways that, historically, have been too cost prohibitive.”
Truby is the June and Donald Brewer Junior Professor of Materials Science and Engineering and Mechanical Engineering at Northwestern’s McCormick School of Engineering, where he directs The Robotic Matter Lab. Taekyoung Kim, a postdoctoral scholar in Truby’s lab and first author on the paper, led the research. Pranav Kaarthik, a Ph.D. candidate in mechanical engineering, also contributed to the work.
While rigid actuators have long been the cornerstone of robot design, their limited flexibility, adaptability and safety have driven roboticists to explore soft actuators as an alternative. To design soft actuators, Truby and his team take inspiration from human muscles, which contract and stiffen simultaneously.
“How do you make materials that can move like a muscle?” Truby asked. “If we can do that, then we can make robots that behave and move like living organisms.”
To develop the new actuator, the team 3D-printed cylindrical structures called “handed shearing auxetics” (HSAs) out of rubber. Difficult to fabricate, HSAs embody a complex structure that enables unique movements and properties. For example, when twisted, HSAs extend and expand. Although Truby and Kaarthik 3D-printed similar HSA structures for robots in the past, they were bound to using expensive printers and rigid plastic resins. As a result, their previous HSAs could not bend or deform easily.
“For this to work, we needed to find a way to make HSAs softer and more durable,” said Kim. “We figured out how to fabricate soft but robust HSAs from rubber using a cheaper and more easily available desktop 3D printer.”
Kim printed the HSAs from thermoplastic polyurethane, a common rubber often used in cellphone cases. While this made the HSAs much softer and more flexible, one challenge remained: how to twist the HSAs to get them to extend and expand. Previous versions of HSA soft actuators used common servo motors to twist the materials into extended and expanded states. But the researchers only achieved successful actuation after assembling two or four HSAs — each with its own motor — together. Building soft actuators in this way presented fabrication and operational challenges. It also reduced the softness of the HSA actuators.
To build an improved soft actuator, the researchers aimed to design a single HSA driven by one servo motor. But first, the team needed to find a way to make a single motor twist a single HSA.
To solve this problem, Kim added a soft, extendable, rubber bellows to the structure that performed like a deformable, rotating shaft. As the motor provided torque — an action that causes an object to rotate — the actuator extended. Simply turning the motor in one direction or the other drives the actuator to extend or contract.
“Essentially, Taekyoung engineered two rubber parts to create muscle-like movements with the turn of a motor,” Truby said. “While the field has made soft actuators in more cumbersome ways, Taekyoung greatly simplified the entire pipeline with 3D printing. Now, we have a practical soft actuator that any roboticist can use and make.”
The bellows added enough support for Kim to build a crawling soft robot from a single actuator that moved on its own. The pushing and pulling motions of the actuator propelled the robot forward through a winding, constrained environment simulating a pipe.
“Our robot can make this extension motion using a single structure,” Kim said. “That makes our actuator more useful because it can be universally integrated into all types of robotic systems.”
The resulting worm-like robot was compact (measuring just 26 centimeters in length) and crawled — both backward and forward — at a speed of just over 32 centimeters per minute. Truby noted that both the robot and artificial bicep become stiffer when the actuator is fully extended. This was yet another property that previous soft robots were unable to achieve.
“Like a muscle, these soft actuators actually stiffen,” Truby said. “If you have ever twisted the lid off a jar, for example, you know your muscles tighten and get stiffer to transmit force. That’s how your muscles help your body do work. This has been an overlooked feature in soft robotics. Many soft actuators get softer when in use, but our flexible actuators get stiffer as they operate.”
Truby and Kim say their new actuator provides yet another step toward more bioinspired robots.
“Robots that can move like living organisms are going to enable us to think about robots performing tasks that conventional robots can’t do,” Truby said.
‘No, Alexa, no!’: designing child-safe AI and protecting children from the risks of the ‘empathy gap’ in large language models
by Nomisha Kurian in Learning, Media and Technology
Artificial intelligence (AI) chatbots have frequently shown signs of an “empathy gap” that puts young users at risk of distress or harm, raising the urgent need for “child-safe AI,” according to a study.
The research, by a University of Cambridge academic, Dr Nomisha Kurian, urges developers and policy actors to prioritise approaches to AI design that take greater account of children’s needs. It provides evidence that children are particularly susceptible to treating chatbots as lifelike, quasi-human confidantes, and that their interactions with the technology can go awry when it fails to respond to their unique needs and vulnerabilities.
The study links that gap in understanding to recent cases in which interactions with AI led to potentially dangerous situations for young users. They include an incident in 2021, when Amazon’s AI voice assistant, Alexa, instructed a 10-year-old to touch a live electrical plug with a coin. Last year, Snapchat’s My AI gave adult researchers posing as a 13-year-old girl tips on how to lose her virginity to a 31-year-old. Both companies responded by implementing safety measures, but the study says there is also a need to be proactive in the long-term to ensure that AI is child-safe. It offers a 28-item framework to help companies, teachers, school leaders, parents, developers and policy actors think systematically about how to keep younger users safe when they “talk” to AI chatbots.
Dr Kurian conducted the research while completing a PhD on child wellbeing at the Faculty of Education, University of Cambridge. She is now based in the Department of Sociology at Cambridge. She argues that AI’s huge potential means there is a need to “innovate responsibly.”
“Children are probably AI’s most overlooked stakeholders,” Dr Kurian said. “Very few developers and companies currently have well-established policies on child-safe AI. That is understandable because people have only recently started using this technology on a large scale for free. But now that they are, rather than having companies self-correct after children have been put at risk, child safety should inform the entire design cycle to lower the risk of dangerous incidents occurring.”
Kurian’s study examined cases where the interactions between AI and children, or adult researchers posing as children, exposed potential risks. It analysed these cases using insights from computer science about how the large language models (LLMs) in conversational generative AI function, alongside evidence about children’s cognitive, social and emotional development.
LLMs have been described as “stochastic parrots”: a reference to the fact that they use statistical probability to mimic language patterns without necessarily understanding them. A similar method underpins how they respond to emotions. This means that even though chatbots have remarkable language abilities, they may handle the abstract, emotional and unpredictable aspects of conversation poorly; a problem that Kurian characterises as their “empathy gap.” They may have particular trouble responding to children, who are still developing linguistically and often use unusual speech patterns or ambiguous phrases. Children are also often more inclined than adults to confide sensitive personal information.
Despite this, children are much more likely than adults to treat chatbots as if they are human. Recent research found that children will disclose more about their own mental health to a friendly-looking robot than to an adult. Kurian’s study suggests that many chatbots’ friendly and lifelike designs similarly encourage children to trust them, even though AI may not understand their feelings or needs.
“Making a chatbot sound human can help the user get more benefits out of it,” Kurian said. “But for a child, it is very hard to draw a rigid, rational boundary between something that sounds human, and the reality that it may not be capable of forming a proper emotional bond.”
Her study suggests that these challenges are evidenced in reported cases such as the Alexa and MyAI incidents, where chatbots made persuasive but potentially harmful suggestions. In the same study in which MyAI advised a (supposed) teenager on how to lose her virginity, researchers were able to obtain tips on hiding alcohol and drugs, and concealing Snapchat conversations from their “parents.” In a separate reported interaction with Microsoft’s Bing chatbot, which was designed to be adolescent-friendly, the AI became aggressive and started gaslighting a user.
Kurian’s study argues that this is potentially confusing and distressing for children, who may actually trust a chatbot as they would a friend. Children’s chatbot use is often informal and poorly monitored. Research by the nonprofit organisation Common Sense Media has found that 50% of students aged 12–18 have used Chat GPT for school, but only 26% of parents are aware of them doing so.
Kurian argues that clear principles for best practice that draw on the science of child development will encourage companies that are potentially more focused on a commercial arms race to dominate the AI market to keep children safe. Her study adds that the empathy gap does not negate the technology’s potential. “AI can be an incredible ally for children when designed with their needs in mind. The question is not about banning AI, but how to make it safe,” she said.
The study proposes a framework of 28 questions to help educators, researchers, policy actors, families and developers evaluate and enhance the safety of new AI tools. For teachers and researchers, these address issues such as how well new chatbots understand and interpret children’s speech patterns; whether they have content filters and built-in monitoring; and whether they encourage children to seek help from a responsible adult on sensitive issues.
The framework urges developers to take a child-centred approach to design, by working closely with educators, child safety experts and young people themselves, throughout the design cycle.
“Assessing these technologies in advance is crucial,” Kurian said. “We cannot just rely on young children to tell us about negative experiences after the fact. A more proactive approach is necessary.”
Microsaccade-inspired event camera for robotics
by Botao He, Ze Wang, Yuan Zhou, Jingxi Chen, Chahat Deep Singh, Haojia Li, Yuman Gao, Shaojie Shen, Kaiwei Wang, Yanjun Cao, Chao Xu, Yiannis Aloimonos, Fei Gao, Cornelia Fermüller in Science Robotics
A team led by University of Maryland computer scientists invented a camera mechanism that improves how robots see and react to the world around them. Inspired by how the human eye works, their innovative camera system mimics the tiny involuntary movements used by the eye to maintain clear and stable vision over time. The team’s prototyping and testing of the camera — called the Artificial Microsaccade-Enhanced Event Camera (AMI-EV).
“Event cameras are a relatively new technology better at tracking moving objects than traditional cameras, but -today’s event cameras struggle to capture sharp, blur-free images when there’s a lot of motion involved,” said the paper’s lead author Botao He, a computer science Ph.D. student at UMD. “It’s a big problem because robots and many other technologies — such as self-driving cars — rely on accurate and timely images to react correctly to a changing environment. So, we asked ourselves: How do humans and animals make sure their vision stays focused on a moving object?”
For He’s team, the answer was microsaccades, small and quick eye movements that involuntarily occur when a person tries to focus their view. Through these minute yet continuous movements, the human eye can keep focus on an object and its visual textures — such as color, depth and shadowing — accurately over time.
“We figured that just like how our eyes need those tiny movements to stay focused, a camera could use a similar principle to capture clear and accurate images without motion-caused blurring,” He said.
The team successfully replicated microsaccades by inserting a rotating prism inside the AMI-EV to redirect light beams captured by the lens. The continuous rotational movement of the prism simulated the movements naturally occurring within a human eye, allowing the camera to stabilize the textures of a recorded object just as a human would. The team then developed software to compensate for the prism’s movement within the AMI-EV to consolidate stable images from the shifting lights. Study co-author Yiannis Aloimonos, a professor of computer science at UMD, views the team’s invention as a big step forward in the realm of robotic vision.
“Our eyes take pictures of the world around us and those pictures are sent to our brain, where the images are analyzed. Perception happens through that process and that’s how we understand the world,” explained Aloimonos, who is also director of the Computer Vision Laboratory at the University of Maryland Institute for Advanced Computer Studies (UMIACS). “When you’re working with robots, replace the eyes with a camera and the brain with a computer. Better cameras mean better perception and reactions for robots.”
The researchers also believe that their innovation could have significant implications beyond robotics and national defense. Scientists working in industries that rely on accurate image capture and shape detection are constantly looking for ways to improve their cameras — and AMI-EV could be the key solution to many of the problems they face.
“With their unique features, event sensors and AMI-EV are poised to take center stage in the realm of smart wearables,” said research scientist Cornelia Fermüller, senior author of the paper. “They have distinct advantages over classical cameras — such as superior performance in extreme lighting conditions, low latency and low power consumption. These features are ideal for virtual reality applications, for example, where a seamless experience and the rapid computations of head and body movements are necessary.”
In early testing, AMI-EV was able to capture and display movement accurately in a variety of contexts, including human pulse detection and rapidly moving shape identification. The researchers also found that AMI-EV could capture motion in tens of thousands of frames per second, outperforming most typically available commercial cameras, which capture 30 to 1000 frames per second on average. This smoother and more realistic depiction of motion could prove to be pivotal in anything from creating more immersive augmented reality experiences and better security monitoring to improving how astronomers capture images in space.
“Our novel camera system can solve many specific problems, like helping a self-driving car figure out what on the road is a human and what isn’t,” Aloimonos said. “As a result, it has many applications that much of the general public already interacts with, like autonomous driving systems or even smartphone cameras. We believe that our novel camera system is paving the way for more advanced and capable systems to come.”
LangNav: Language as a Perceptual Representation for Navigation
by Bowen Pan, Rameswar Panda, SouYoung Jin, Rogerio Feris, Aude Oliva, Phillip Isola, Yoon Kim in Submitted to arXiv
Someday, you may want your home robot to carry a load of dirty clothes downstairs and deposit them in the washing machine in the far-left corner of the basement. The robot will need to combine your instructions with its visual observations to determine the steps it should take to complete this task.
For an AI agent, this is easier said than done. Current approaches often utilize multiple hand-crafted machine-learning models to tackle different parts of the task, which require a great deal of human effort and expertise to build. These methods, which use visual representations to directly make navigation decisions, demand massive amounts of visual data for training, which are often hard to come by.
To overcome these challenges, researchers from MIT and the MIT-IBM Watson AI Lab devised a navigation method that converts visual representations into pieces of language, which are then fed into one large language model that achieves all parts of the multistep navigation task.
Rather than encoding visual features from images of a robot’s surroundings as visual representations, which is computationally intensive, their method creates text captions that describe the robot’s point-of-view. A large language model uses the captions to predict the actions a robot should take to fulfill a user’s language-based instructions.
Because their method utilizes purely language-based representations, they can use a large language model to efficiently generate a huge amount of synthetic training data. While this approach does not outperform techniques that use visual features, it performs well in situations that lack enough visual data for training. The researchers found that combining their language-based inputs with visual signals leads to better navigation performance.
“By purely using language as the perceptual representation, ours is a more straightforward approach. Since all the inputs can be encoded as language, we can generate a human-understandable trajectory,” says Bowen Pan, an electrical engineering and computer science (EECS) graduate student and lead author of a paper on this approach.
Since large language models are the most powerful machine-learning models available, the researchers sought to incorporate them into the complex task known as vision-and-language navigation, Pan says. But such models take text-based inputs and can’t process visual data from a robot’s camera. So, the team needed to find a way to use language instead. Their technique utilizes a simple captioning model to obtain text descriptions of a robot’s visual observations. These captions are combined with language-based instructions and fed into a large language model, which decides what navigation step the robot should take next.
The large language model outputs a caption of the scene the robot should see after completing that step. This is used to update the trajectory history so the robot can keep track of where it has been. The model repeats these processes to generate a trajectory that guides the robot to its goal, one step at a time. To streamline the process, the researchers designed templates so observation information is presented to the model in a standard form — as a series of choices the robot can make based on its surroundings. For instance, a caption might say “to your 30-degree left is a door with a potted plant beside it, to your back is a small office with a desk and a computer,” etc. The model chooses whether the robot should move toward the door or the office.
“One of the biggest challenges was figuring out how to encode this kind of information into language in a proper way to make the agent understand what the task is and how they should respond,” Pan says.
When they tested this approach, while it could not outperform vision-based techniques, they found that it offered several advantages. First, because text requires fewer computational resources to synthesize than complex image data, their method can be used to rapidly generate synthetic training data. In one test, they generated 10,000 synthetic trajectories based on 10 real-world, visual trajectories.
The technique can also bridge the gap that can prevent an agent trained with a simulated environment from performing well in the real world. This gap often occurs because computer-generated images can appear quite different from real-world scenes due to elements like lighting or color. But language that describes a synthetic versus a real image would be much harder to tell apart, Pan says. Also, the representations their model uses are easier for a human to understand because they are written in natural language.
“If the agent fails to reach its goal, we can more easily determine where it failed and why it failed. Maybe the history information is not clear enough or the observation ignores some important details,” Pan says.
In addition, their method could be applied more easily to varied tasks and environments because it uses only one type of input. As long as data can be encoded as language, they can use the same model without making any modifications. But one disadvantage is that their method naturally loses some information that would be captured by vision-based models, such as depth information. However, the researchers were surprised to see that combining language-based representations with vision-based methods improves an agent’s ability to navigate.
“Maybe this means that language can capture some higher-level information than cannot be captured with pure vision features,” he says.
This is one area the researchers want to continue exploring. They also want to develop a navigation-oriented captioner that could boost the method’s performance. In addition, they want to probe the ability of large language models to exhibit spatial awareness and see how this could aid language-based navigation.
CARMEN: A Cognitively Assistive Robot for Personalized Neurorehabilitation at Home
by Anya Bouzida, Alyssa Kubota, Dagoberto Cruz-Sandoval, Elizabeth W. Twamley, Laurel D. Riek in Proceedings of the 2024 ACM/IEEE International Conference on Human-Robot Interaction (HRI ‘24)
Meet CARMEN, short for Cognitively Assistive Robot for Motivation and Neurorehabilitation-a small, tabletop robot designed to help people with mild cognitive impairment (MCI) learn skills to improve memory, attention, and executive functioning at home.
Unlike other robots in this space, CARMEN was developed by the research team at the University of California San Diego in collaboration with clinicians, people with MCI, and their care partners. To the best of the researchers’ knowledge, CARMEN is also the only robot that teaches compensatory cognitive strategies to help improve memory and executive function.
“We wanted to make sure we were providing meaningful and practical inventions,” said Laurel Riek, a professor of computer science and emergency medicine at UC San Diego and the work’s senior author.
MCI is an in-between stage between typical aging and dementia. It affects various areas of cognitive functioning, including memory, attention, and executive functioning. About 20% of individuals over 65 have the condition, with up to 15% transitioning to dementia each year. Existing pharmacological treatments have not been able to slow or prevent this evolution, but behavioral treatments can help.
Researchers programmed CARMEN to deliver a series of simple cognitive training exercises. For example, the robot can teach participants to create routine places to leave important objects, such as keys; or learn note taking strategies to remember important things. CARMEN does this through interactive games and activities.
The research team designed CARMEN with a clear set of criteria in mind. It is important that people can use the robot independently, without clinician or researcher supervision. For this reason, CARMEN had to be plug and play, without many moving parts that require maintenance. The robot also has to be able to function with limited access to the internet, as many people do not have access to reliable connectivity. CARMEN needs to be able to function over a long period of time. The robot also has to be able to communicate clearly with users; express compassion and empathy for a person’s situation; and provide breaks after challenging tasks to help sustain engagement.
Researchers deployed CARMEN for a week in the homes of several people with MCI, who then engaged in multiple tasks with the robot, such as identifying routine places to leave household items so they don’t get lost, and placing tasks on a calendar so they won’t be forgotten. Researchers also deployed the robot in the homes of several clinicians with experience working with people with MCI. Both groups of participants completed questionnaires and interviews before and after the week-long deployments.
After the week with CARMEN, participants with MCI reported trying strategies and behaviors that they previously had written off as impossible. All participants reported that using the robot was easy. Two out of the three participants found the activities easy to understand, but one of the users struggled. All said they wanted more interaction with the robot.
“We found that CARMEN gave participants confidence to use cognitive strategies in their everyday life, and participants saw opportunities for CARMEN to exhibit greater levels of autonomy or be used for other applications,” the researchers write.
Next steps include deploying the robot in a larger number of homes. Researchers also plan to give CARMEN the ability to have conversations with users, with an emphasis on preserving privacy when these conversations happen. This is both an accessibility issue (as some users might not have the fine motor skills necessary to interact with CARMEN’s touch screen), as well as because most people expect to be able to have conversations with systems in their homes. At the same time, researchers want to limit how much information CARMEN can give users. “We want to be mindful that the user still needs to do the bulk of the work, so the robot can only assist and not give too many hints,” Riek said.
Researchers are also exploring how CARMEN could assist users with other conditions, such as ADHD.
Subscribe to Paradigm!
Medium. Twitter. Telegram. Telegram Chat. Reddit. LinkedIn.
Main sources
Research articles
