Ever wanted to hug a robot? It sounds unappealing… but we’re working on it. Soft robotics are robots that contain parts that are soft or flexible. It’s a rapidly expanding frontier research field. Within this space, we are designing robots that can interact safely and intuitively with people and the environment.
Soft robotics has a broad range of industrial applications. For example, in agriculture a delicate touch may be required to pick fruit without bruising it. In medicine, soft robotic parts offer the adaptability, robustness and safety required to interact with the human body.
While this is all exciting in theory, in reality, soft robotics are still largely confined to the lab. We’re trying to change that and get these emerging technologies into the field to help solve our greatest challenges.
Recently, we presented four cutting-edge papers at the 6th IEEE-RAS International Conference on Soft Robotics (RoboSoft). Here are our latest innovations.
You may know grains as a staple of a healthy diet, but they’re also a key ingredient in many soft robotic parts. Granular jamming is one technique for creating soft robotic grippers.
Granular jamming works by packing grains (e.g. coffee) into a membrane (e.g. a balloon) and attaching a vacuum line. Under atmospheric pressure, the grains will move and deform around an object like a fluid. When the air is suctioned out, the particles jam together and form a strong grasp on the object.
But not all grains are made equal. At Robosoft, our researchers publicly released the largest-ever dataset of grain types for use in granular jamming.
Dr David Howard is the AI4Design portfolio lead, as well as leading the Soft Robotics Cluster in the Robotics and Autonomous Systems Group at CSIRO’s Data61. He said the data will provide a baseline for developers of soft robotics, allowing them to tailor their grain to the task at hand.
"We found that quite soft, rubbery grains are the best for gripping. However, say you’ve got a robot out in the field that’s likely to get tossed around a bit. You need to protect the electronics using a skin. We found large, hard grains performed better for shock absorption," David said.
But it’s not just the size and shape of the grain, it’s what you do with it. Our researchers also recently discovered vibrations from sound waves can be used to enhance granular jamming performance.
"You can tune the behaviour of these kinds of grippers using vibrations, resulting in stronger grips. What’s more, this improvement in performance is affordable; all you need is a single 3D-printed adapter and a $20 amplifier," he said.
These new discoveries allow developers to optimise soft robotic solutions, taking us one step closer to deploying them in the field.
This innovation offers a sweet solution to a sticky problem. While granular jammers allow robots to pick up objects of various shapes securely, they rely on the object resting against a flat surface.
The Jamming Donut works by combining an outer ring which inflates, with an inner granular jammer to grasp objects suspended in air.
This technology is an innovative design that outperforms other systems of grippers due to its unique combination of components. These limit the possibility of robot's having the equivalent of butterfingers.
Again, it has the potential for a wide range of real-world applications including picking up soft and hard produce, and opening doorknobs – a task that other grippers struggle to perform well.
Design at the click of a button
Before a robot with soft components enters the real world, you need to design and build it. Our AI4Design portfolio is developing new methods of designing complicated, integrated robotic systems with the help of artificial intelligence. They are applying the same techniques to create next-generation industrial equipment to support our Missions.
David's team have been working on automating the design process of robot grippers using AI techniques. They have created software that uses AI to automatically generate fit-for-purpose pneumatic grippers within material and size constraints inputted by users.
This work is one of the fundamental building blocks needed in order to realise the dream of autonomous design, saving time and resources and augmenting human creativity.
“We’re looking at using generative AI to creatively explore the design space, then using physics-based modelling and simulation coupled with real-world experimentation to provide some ground truth data,” said David.
“This can then be used as a tool to do rapid, blanket exploration of these really complicated, interesting design spaces to find intuitive, novel, high performance designs.”
The Soft Robotics Cluster is supported by the Robotics and Autonomous Systems Group, Cyber-Physical Systems Program Data61. AI4 Design is co-funded by the AI for Missions initiative and the Future Digital Manufacturing Fund. Find out more about our soft robotic capability by contacting David.