Enabling complexity in fluidically actuated soft robots via onboard control hardware
Embargo Date
2025-05-30
OA Version
Citation
Abstract
Soft robots offer advantages over traditional rigid robots with regards to compliance, dexterity, and safety, making them appropriate for use in delicate and unstructured environments. Fluidically actuated soft robots, whereby a pressurized fluid cause the robot to bend, extend, or otherwise deform, are among the most popular due to their small size and high power density. Such actuators typically require a pressure regulation system outside of the robot itself to control the inflation of the actuators. Complex soft robots use many actuators to conduct tasks like locomotion and object manipulation, historically burdening the robot with many tubes tethering it to its pressure controller. In an effort to reduce the number of tubes required to control a soft robot, many categories of onboard pressure control hardware been developed. These can be divided into two primary categories, those that allow for electronic control and those that are strictly mechanical. While strictly mechanical systems can be advantageous in remote locations or in other situations where electricity is unavailable or unreliable, electronically controlled systems are unmatched with regards to their ability to adapt and respond to their environment in real time. However, electronically controlled hardware for soft robots has remained limited to date with regards to size, number of controllable DoFs, and maximum pressure controllable. Therefore, in this dissertation we present novel methods of introducing electronic control hardware onboard fluidically actuated soft robots.
First, we present a method of controlling soft robots using magnetorheological (MR) fluids. MR fluids are a class of smart fluid that consist of iron particles in a carrier fluid like water along with stability enhancing additives. In the presence of a magnetic field, MR fluids solidify. This effect is quantified as an increase in yield stress and viscosity, and the effect intensifies with increasing magnetic field strength. We demonstrate a method whereby permanent magnets were used to solidify a continuously flowing MR fluid, resulting in an increase in flow pressure and subsequently the inflation of soft actuators. Up to five soft actuators are controlled with a single inlet and single outlet.
Next, the method of controlling soft robots with MR fluids was improved via the introduction of electropermanent magnets (EPMs) as an electronically controlled magnetic field source. EPMs are an assembly of one neodymium permanent magnet, one alnico permanent magnet, an electromagnet, and steel end caps. The two permanent magnets have similar remanence, but the alnico magnet has a lower coercivity, making it susceptible to changes in the magnitude and orientation of its magnetic field. A pulse of current with duration on the order of several hundred microseconds reorients the magnet field of the alnico magnet component, allowing the field of the overall assembly to be controlled from zero up to a maximum value dictated by the permanent magnets' geometry and material grade. We use EPMs to modulate the magnetic field exerted on an MR fluid flowing through a soft robot. With the flow channel passing between the EPM's end caps, the magnetic field is applied orthogonally to MR fluid's flow. By adjusting the length of the current pulse through the EPM's electromagnet, we control the bending angle and force output of several classes of common soft fluidic actuators. We then control multi-DoF systems of actuators connected in series and parallel, demonstrating real-time control over the motion of such robots.
Finally, we introduce a novel EPM valve for the control of pneumatic soft robots. This valve does not rely on MR fluid, but instead uses the force exerted at the EPM's end cap on a steel plate. A thin flow channel manufactured from 70 µm thick thermoplastic elastomer (TPE) film is placed directly under the end cap such that with the EPM magnetized, the force clamps the channel shut, preventing the flow of air. The valve is shown to block up to 30 kPa of flow pressure. We demonstrate the valve's use for controlling a three DoF stacked balloon actuator (SBA) and a stack of two SBAs connected end-to-end. The two SBAs are then combined side-by-side to create a robot that can grasp objects or locomote without modifying its hardware. In this way we show a system wherein onboard electronic components enable a soft robot to be reprogrammed to complete different tasks and interact with its environment.
These efforts show important progress toward the development of soft robots that can perform independently. Bringing control hardware onboard the robot itself reduces the impact of bulky pressure tubes on the robot's size and mobility. This in turn brings soft robots closer to operating in a truly untethered, autonomous manner. Retaining electronic control furthermore allows the robot to integrate sensors and algorithms to respond to obstacles and objects of interest in its environment. This is an essential aspect of bringing together soft robot hardware and software methods developed for rigid robots, including machine learning. Such integration will permit soft robots to reach their potential to solve the critical challenges facing robotics today.