Robotics & AI Updates vol.100

December 19th 2024

Check out latest research updates in the field

TL;DR

New breakthrough helps free up space for robots to ‘think’
A gentle and versatile robotic gripper for efficient crop harvesting
Ultra-sensitive robotic ‘finger’ can take patient pulses, check for lumps
Simulation mimics how the brain grows neurons, paving the way for future disease treatments
New AI models of plasma heating lead to important corrections in computer code used for fusion research

Latest News & Research

Frequency‐Controlled Fluidic Oscillators for Soft Robots

by Mostafa Mousa, Ashkan Rezanejad, Benjamin Gorissen, Antonio E. Forte in Advanced Science

Engineers have worked out how to give robots complex instructions without electricity for the first time which could free up more space in the robotic ‘brain’ for them to ‘think’.

Mimicking how some parts of the human body work, researchers from King’s College London have transmitted a series of commands to devices with a new kind of compact circuit, using variations in pressure from a fluid inside it.

They say this world first opens up the possibility of a new generation of robots, whose bodies could operate independently of their built-in control centre, with this space potentially being used instead for more complex AI powered software.

“Delegating tasks to different parts of the body frees up computational space for robots to ‘think,’ allowing future generations of robots to be more aware of their social context or even more dexterous. This opens the door for a new kind of robotics in places like social care and manufacturing,” said Dr Antonio Forte, Senior Lecturer in Engineering at King’s College London and senior author of the study.

Reconfigurable valve design and characterization.

The findings could also enable the creation of robots able to operate in situations where electricity-powered devices cannot work, such as exploration in irradiated areas like Chernobyl which destroy circuits, and in electric sensitive environments like MRI rooms. The researchers also hope that these robots could eventually be used in low-income countries which do not have reliable access to electricity.

Dr Forte said: “Put simply, robots are split into two parts: the brain and the body. An AI brain can help run the traffic system of a city, but many robots still struggle to open a door — why is that?
“Software has advanced rapidly in recent years, but hardware has not kept up. By creating a hardware system independent from the software running it, we can offload a lot of the computational load onto the hardware, in the same way your brain doesn’t need to tell your heart to beat.”

Currently, all robots rely on electricity and computer chips to function. A robotic ‘brain’ of algorithms and software translates information to the body or hardware through an encoder, which then performs an action.

In ‘soft robotics,’ a field which creates devices like robotic muscles out of soft materials, this is particularly an issue as it introduces hard electronic encoders and puts strain on the software for the material to act in a complex way, e.g. grabbing a door handle.

To circumvent this, the team developed a reconfigurable circuit with an adjustable valve to be placed within a robot’s hardware. This valve acts like a transistor in a normal circuit and engineers can send signals directly to hardware using pressure, mimicking binary code, allowing the robot to perform complex manoeuvres without the need for electricity or instruction from the central brain. This allows for a greater level of control than current fluid-based circuits.

By offloading the work of the software onto the hardware, the new circuit frees up computational space for future robotic systems to be more adaptive, complex, and useful.

As a next step, the researchers now hope to scale up their circuits from experimental hoppers and pipettes and embed them in larger robots, from crawlers used to monitor power plants to wheeled robots with entirely soft engines.

Mostafa Mousa, Post-graduate Researcher at King’s College London and author, said: “Ultimately, without investment in embodied intelligence robots will plateau. Soon, if we do not offload the computational load that modern-day robots take on, algorithmic improvements will have little impact on their performance. Our work is just a first step on this path, but the future holds smarter robots with smarter bodies.”

Soft yet secure: Exploring membrane buckling for achieving a versatile grasp with a rotation-driven squeezing gripper

by Khoi Thanh Nguyen, Nhan Huu Nguyen, Van Anh Ho in The International Journal of Robotics Research

Robotic grippers have become essential across many industries, including manufacturing, packaging, and logistics, mainly for pick-and-place tasks. Recently, the demand for robotic grippers has also expanded into agriculture, where they are used for harvesting and packaging tasks. However, conventional robotic grippers struggle with the unique shapes, properties, and delicate nature of different crops. Consequently, there has been an increasing demand for more versatile robots that can adapt to objects with various shapes, sizes, and textures.

Robotic grippers that are made of soft materials have emerged as a potential solution to the above problem. However, current methods for adapting these grippers to complex geometries rely on complex control and planning generated by data-based models. These models require a large amount of data, limiting their general applicability. Additionally, integrating a sensory system into their soft body requires complex designs and sophisticated fabrication methods.

To this end, a team of researchers from the Japan Advanced Institute of Science and Technology (JAIST), led by Associate Professor Van Anh Ho, along with Assistant Professor Nguyen Huu Nhan and doctoral course student Nguyen Thanh Khoi, developed an innovative soft robotic gripper named ROtation-based Squeezing grippEr or ROSE.

“ROSE takes inspiration from the blooming states of a rose to generate grasping action. It offers a simpler approach to real-farm harvesting by gently grasping objects using a unique “wrinkling” phenomenon. Unlike conventional grippers, ROSE doesn’t require complex control and planning strategies to adapt to various agricultural products with diverse shapes, sizes, and textures,” explainsProf. Ho. They also employed a simulation model to fully understand and optimize the grasping mechanism of ROSE. This study was published in a special issue, RSS2023, of The International Journal of Robotics Research.

ROSE consists of an isolated cup-shaped chamber formed by two thin, soft elastomer layers, with a separation between the interior and outer layers. Rotating only the inner layer using an external motor produces a deformation in the layers. Specifically, this twisting motion of the inner layer results in a strain mismatch between the outer and inner layers, resulting in the formation of a series of wrinkle-like inward folds, a process termed ‘Wrinkling’. This unique mechanism shrinks the central space in ROSE, which allows it to gently grasp any object present within this central space.

To refine this mechanism, the researchers studied the ‘wrinkling’ process through a finite element method-based simulation model. The simulations revealed a correlation between different geometrical features, including thickness, diameter, and height. Notably, they found that an appropriate distribution of ROSE’s skin thickness, that is, the separation between the layers, has a significant influence on its grasping performance. As a result, the researchers tested two different thickness distribution strategies, namely linear and non-linear distribution, which significantly improved ROSE’s grasping performance compared to a constant thickness. Moreover, the simulations also highlighted the importance of the ratio between the gripper’s diameter and height. The simulation results were validated through various experiments, verifying ROSE’s capability to carry out tasks that are difficult for traditional grippers.

Furthermore, the researchers demonstrated the practical applications of ROSE in agriculture by using it to harvest strawberries and mushrooms. ROSE achieved high success rates in picking up these crops in multiple trials, regardless of whether they were stiff or soft. It also succeeded in picking up a clump of mushrooms without breaking any piece, provided the clump size fit within the grasping space.

“ROSE is one of the first grippers to utilize buckling deformation as a gripping method, challenging the conventional mindset that the buckling phenomenon is an undesired feature. Moreover, the practical application of ROSE in agricultural settings is a game-changer for real-farm harvesting. ROSE’s ability to adapt to varying textures and shapes makes it highly effective in these tasks. This not only improves efficiency but can also address the growing labor shortages in agriculture, particularly in regions with aging populations,” highlights Prof. Ho emphasizing the potential of ROSE.

Toward human-like touch sense via a bioinspired soft finger with self-decoupled bending and force sensing

by Yufeng Wang, Houping Wu, Tonglin Li, Jinxing Wang, Zhipeng Wei, Hongbo Wang in Cell Reports Physical Science

Researchers at the University of Science and Technology of China have developed a soft robotic “finger” with a sophisticated sense of touch that can perform routine doctor office examinations, including taking a patient’s pulse and checking for abnormal lumps.

Such technology could make it easier for doctors to detect diseases such as breast cancer early on, when they are more treatable. It may also help patients feel at ease during physical exams that can seem uncomfortable and invasive.

“By further development to improve its efficiency, we also believe that a dexterous hand made of such fingers can act as a ‘Robodoctor’ in a future hospital, like a physician,” says Hongbo Wang, a sensing technologies researcher at the University of Science and Technology of China and an author of the study. “Combined with machine learning, automatic robotic examination and diagnosis can be achieved, particularly beneficial for these undeveloped areas where there is a serious shortage in health workers.”

While rigid robotic fingers already exist, experts have raised concerns that these devices might not be up to the delicate tasks required in a doctor’s office setting. Some have pointed to potential safety issues, including a fear that overzealous robotic fingers could rupture lumps during examinations. More recently, scientists have developed lightweight, safe, and low-cost soft robotics that can recreate the movements of human hands. However, these devices haven’t been able to sense the complex properties of objects they touch the way real fingers do.

“Despite the remarkable progress in the last decade, most soft fingers presented in the literature still have substantial gaps compared to human hands,” the authors write, noting that robotic fingers have not been ready to handle “‘real world’ scenarios.”

To overcome this challenge, the researchers developed a simple device that contains conductive fiber coils with two parts — a coil wound on each air chamber of the device’s bending actuators (the parts that enable it to move) and a twisted liquid metal fiber mounted at the fingertip. By measuring properties that affect how the device’s electrical current flows, the team found that they could monitor, in real time, how far the finger bends as it touches an object and the force at the fingertip. In this way, the device could perceive an object’s properties as effectively as human touch.

To test the device, the researchers started by brushing a feather against its fingertip.

“The magnified view clearly shows the resistance change, indicating its high sensitivity in force sensing,” the authors write.

Next, they tapped and pushed the fingertip with a glass rod and repeatedly bent the finger, observing that the device’s sensors accurately perceived the type and quantity of force they applied. To test the finger’s medical chops, they mounted it on a robotic arm and watched as it identified three lumps embedded in a large silicone sheet, pressing on them like a doctor would. While mounted on the robotic arm, the finger also correctly located an artery on a participant’s wrist and took their pulse.

“Humans can easily recognize the stiffness of diverse objects by simply pressing it with their finger,” the authors write. “Similarly, since the [device] has the ability to sense both its bending deformation and the force at the fingertip, it can detect stiffness similar to our human hand by simply pressing an object.”

In addition to taking pulses and examining simulated lumps, the researchers found that the robotic finger can type “like a human hand,” spelling out the word “hello.”

By using additional sensors to create even more flexibility in the robotic finger’s joints, allowing the device to move in multiple directions like a human finger, it may be ready to perform effective and efficient medical examinations in the near future, the authors conclude.

“We hope to develop an intelligent, dexterous hand, together with a sensorized artificial muscle-driven robotic arm, to mimic the unparalleled functions and fine manipulations of the human hands,” said Wang.

Calibration of stochastic, agent-based neuron growth models with approximate Bayesian computation

by Tobias Duswald, Lukas Breitwieser, Thomas Thorne, Barbara Wohlmuth, Roman Bauer in Journal of Mathematical Biology

A new computer simulation of how our brains develop and grow neurons has been built by scientists from the University of Surrey. Along with improving our understanding of how the brain works, researchers hope that the models will contribute to neurodegenerative disease research and, someday, stem cell research that helps regenerate brain tissue.

The research team used a technique called Approximate Bayesian Computation (ABC), which helps fine-tune the model by comparing the simulation with real neuron growth. This process ensures that the artificial brain accurately reflects how neurons grow and form connections in real life.

The simulation was tested using neurons from the hippocampus — a critical region of the brain involved in memory retention. The team found that their system successfully mimicked the growth patterns of real hippocampal neurons, showing the potential of this technology to simulate brain development in fine detail.

Dr Roman Bauer from the University of Surrey’s School of Computer Science and Electronic Engineering said: “How our brain works is still one of the greatest mysteries in science. With this simulation, and the rapid advancements in artificial intelligence, we’re getting closer to understanding how neurons grow and communicate. We hope that one day this work could lead to better treatments for devastating diseases like Alzheimer’s or Parkinson’s — changing lives for millions.”

The accuracy of the model is closely tied to the quality of the data used to calibrate it. If the real-life neuron data is limited or incomplete, the precision of the simulation may decrease.

While the current model has shown impressive results in replicating the growth of specific neurons, such as hippocampal pyramidal cells, further adjustments may be needed to accurately simulate other types of neurons or regions of the brain.

The computer simulation is built from the BioDynaMo software, which Dr Bauer co-developed. The software supports scientists to easily create, run, and visualise multi-dimensional agent-based simulations, be they biological, sociological, ecological or financial.

Real-time capable modeling of ICRF heating on NSTX and WEST via machine learning approaches

by Á. Sánchez-Villar, Z. Bai, N. Bertelli, E.W. Bethel, J. Hillairet, T. Perciano, S. Shiraiwa, G.M. Wallace, J.C. Wright in Nuclear Fusion

New AI models for plasma heating can do more than was previously thought possible, not only increasing the prediction speed 10 million times while preserving accuracy, but also correctly predicting plasma heating in cases where the original numerical code failed.

“With our intelligence, we can train the AI to go even beyond the limitations of available numerical models,” said Álvaro Sánchez-Villar, an associate research physicist at the U.S. Department of Energy’s (DOE) Princeton Plasma Physics Laboratory (PPPL). Sánchez-Villar is the lead author of a new peer-reviewed journal article about the work. It was part of a project that spanned five research institutions.

The models use machine learning, a type of AI, to try to predict the way electrons and ions in a plasma behave when ion cyclotron range of frequency (ICRF) heating is applied in fusion experiments. The models are trained on data generated by a computer code. While much of the data agreed with past results, in some extreme scenarios the data wasn’t what they expected.

Heating profiles for deuterium are shown in (d) minor, (e) major and (f) critical outlier cases. In black, the original numerical code is shown with outlier features (spikes).

“We observed a parametric regime in which the heating profiles featured erratic spikes in rather arbitrary locations,” said Sánchez-Villar. “There was nothing physical to explain those spikes.”

New AI models for plasma heating can do more than was previously thought possible, not only increasing the prediction speed 10 million times while preserving accuracy, but also correctly predicting plasma heating in cases where the original numerical code failed. The models will be presented on October 11 at the 66th Annual Meeting of the American Physical Society Division of Plasma Physics in Atlanta.

“With our intelligence, we can train the AI to go even beyond the limitations of available numerical models,” said Álvaro Sánchez-Villar, an associate research physicist at the U.S. Department of Energy’s (DOE) Princeton Plasma Physics Laboratory (PPPL). Sánchez-Villar is the lead author on a new peer-reviewed journal article in Nuclear Fusion about the work. It was part of a project that spanned five research institutions.

“We observed a parametric regime in which the heating profiles featured erratic spikes in rather arbitrary locations,” said Sánchez-Villar. “There was nothing physical to explain those spikes.”
“This means that, practically, our surrogate implementation was equivalent to fixing the original code, just based on a careful curation of the data,” said Sánchez-Villar. “As with every technology, with an intelligent use, AI can help us solve problems not only faster, but better than before, and overcome our own human constraints.”

As expected, the models also improved the computation times for ICRF heating. Those times fell from roughly 60 seconds to 2 microseconds, enabling faster simulations without notably impacting the accuracy. This improvement will help scientists and engineers explore the best ways to make fusion a practical power source.