TU alumnus Bas Overvelde has developed controllably inflatable elements suitable to be used as soft muscles for robots. He published his results, achieved at Harvard University, in PNAS last Monday.
Overvelde is a PhD student in the Bertoldi Group at the School of Engineering and Applied Sciences at Harvard. On this project, two students from Delft assisted him: Tamara Kloek (Applied Mathematics) and Jonas D’haen (Aerospace Engineering).
Their project works towards the development of the next generation of soft robots – machines that work with inflatable elements instead of electric motors and gears. In collaboration with the Harvard Biodesign Lab Overvelde previously worked on applications in prostheses, such as an artificial heart muscle or sensing and control devices.
The problem with these inflatable actuators is their linearity: to achieve large deformations or high forces, large amounts of fluids need to be pumped into them. Therefore, such systems are slow, or rely on powerful pumps.
Not anymore, said Overvelde. In his article, he makes use of inherent non-linear behaviour of inflatable elements, that, when combined, achieve sudden changes in length, shape, internal pressure and exerted force.
The simplest illustration of this non-linear behaviour is a popular high-school demonstration. You fully inflate one balloon, and another one halfway. You connect them with a tube and have the audience predict what will happen. Often, they are in for a surprise: instead of the balloons equalizing in size, all the internal volume flows to the bigger balloon.
Overvelde used inflatable latex tubes instead of balloons. He was able to control their individual non-linear behaviour by surrounding them with plastic cages, called braids. Once a ballooning tube hits the wall of its cage, non-linearity sets in because the volume cannot expand further. The properties of such an element can be tuned by varying the lengths of the tube and the braids (the latter is always longer to start with).
Overvelde has set up a mathematical model that describes the behaviour of such soft muscles. The computer model allows him to assemble actuators by only using the measured data of the single elements. By using this approach, from the 36 fabricated individual elements, 630 actuators consisting of 2 elements could be assembled. Each of these ‘combined’ actuators shows a different response upon inflation.
For example, in one actuator a large internal volume flow is triggered while only inserting a small amount of fluid.
Another example that Overvelde describes is the coupling of multiple fine-tuned muscles to perform a sequence of motions autonomously, such as walking. He has already shown that he can accelerate the speed of soft actuators more than a tenfold by using air instead of water and adding an extra reservoir.
Comments are closed.