In the world of assistive devices and soft robotics, soft actuators are becoming the stars. These actuators have very fast response times and a large output power density, making them highly desirable. But despite all the promise of this technology, adoption rates have been quite low. This is because design and fabrication of soft actuators has remained a difficult challenge, requiring a good deal of slow, labor-intensive manual work. To add further utility to the actuators, the next logical step is to incorporate sensors; however, simple and established methods for doing so do not exist, so again implementations of such designs are relatively few and far between, with no practical means for production at scale.
Soft actuators may soon become more practical to produce thanks to research conducted by a team at MIT that has described a manufacturing process that involves knitting. I know what you are thinking, and no, smart aleck, this research was not conducted by the MIT Class of 1965 Alumni Club in between making socks and sweaters for their grandchildren. Rather, a method was developed that allows for sensorized, fabric articles to be designed computationally and produced via commercial knitting machines. The fabric is designed to trigger movement along predetermined paths through the inflation of elastomeric silicone tubes that are inserted into the designs.
The actuation response of the fabric is directed by strategically placing elastic stitches among the normal stitches during the machine-knitting procedure. The additional stiffness added by the elastic stitches forces the silicone tube to bend in a desired direction and allows the actuator to perform useful work. An optional, secondary layer of conductive thread can also be incorporated into the fabric. This thread can be used to produce sensors, and the team has demonstrated both pressure and swept frequency capacitive sensors in action using their technique by carefully laying out the pattern of this secondary layer.
This newly introduced fabrication method for soft actuators creates inexpensive devices using long-established manufacturing processes. These processes already allow a wide variety of materials, with varying properties, to be incorporated into the fabrics at the stitch level. This feature allows many types of functionalities to be programmed into the product without changing the manufacturing methods. Entire wearable devices can be produced within minutes, and there is a large degree of flexibility in what can be produced, from wearable assistive devices to soft robotics components. Being naturally lightweight and breathable by their very makeup, wearable devices in particular look to be a promising application of this process.
The researchers built a number of gadgets that showcase the capabilities of their fabrication technique. In one case, they built a soft gripper with six actuators. There is one actuator per finger, and two for the thumb, with the actuators being attached to a 3D printed, palm-shaped base. By controlling which actuators are inflated, and when, the team demonstrated that the gripper is capable of grasping a wide variety of objects, including tennis balls and wooden blocks. In a related demonstration, they built a two-fingered gripper that also had integrated pressure and swept frequency capacitive sensors in the fingers that provided real time feedback about the quality of the grasp that the gripper had on objects. Both of these demonstrations highlight possible applications in industrial, and other areas of, soft robotics.
Assistive technologies were also demonstrated in the form of a wearable glove with five actuators. One actuator is positioned on top of each finger, allowing the device to give a boost in gripping objects for those that may have a motor or neurological disability that makes this difficult. In another case, an actuator was positioned in the sleeve of a shirt, which then provided help with elbow bending. Electromyography was performed on study participants using these devices, and it was found that they only needed to exert a small amount of muscular effort in their activities.
Currently, this method is limited by the fact that bending motions are only possible in a single direction at a time. The team also recognizes the need to improve their computational design tools such that a user only needs to specify their desired end state of actuation, rather than specifying all of the design details that are required to achieve that end state. Despite these limitations, the flexibility, low-cost, and short design iteration time that this technique offers is sure to move the ball forward for soft actuation devices.