Intern Report: Development of a Faster Shape Memory Alloy Actuator
Hi, I am Ken Suzuki, a third-year undergraduate mechanical engineering student from the University of Massachusetts Amherst, located in the United States. I joined the internship program at OMRON SINIC X (OSX) in Tokyo, Japan, from June 1st through August 31st of 2023.
At OSX, I was involved in the development of a shape memory alloy (SMA) actuator, under the supervision of Dr. Kazutoshi Tanaka and Dr. Masashi Hamaya of OMRON SINIC X. I would like to express my tremendous gratitude to Dr. Kazutoshi Tanaka Dr. Masashi Hamaya for giving me the opportunity to be a part of the internship program at OMRON SINIC X. I am very grateful for their invaluable advice and supervision, and their support allowed me to grow as a mechanical engineering student and aspiring roboticist.
In this report, I will share what I worked on during my three-month internship period, including my design process and background information about shape memory actuators from various research papers and other reputable online sources. In the end, I will also share my thoughts about interning at OMRON SINIC X.
How Shape Memory Alloys Work
A shape memory alloy, or an SMA for short, is a special type of alloy that is characterized by its ability to demonstrate a “shape memory effect” [9]. This effect is caused by phase transformations in the crystal structure of the alloy that allow it to transform into any memorized shape [9]. In the case of our project, we utilized Nitinol (NiTi) wire. Nitinol wire has a special property, where the wire contracts when it is heated, and then transforms back to its original length when it is cooled down [9].
Advantages/Disadvantages of Using Shape Memory Alloys
SMA actuators have favorable advantages compared to traditional methods of actuation: higher power density (high output to weight ratio), compact size, silent actuation, and lightweight form factor [1]. However, SMA actuators also hold many disadvantages. When actuated, SMA wires exhibit nonlinear behavior when comparing their linear displacement with temperature [3]. They are also subject to hysteresis effects, where transformation temperatures between different phases depend on the history of the transformation [9]. This makes SMA actuators difficult to control.
SMAs come in different shapes in sizes, but typically come in the form of thin wire. SMA wires mainly come in two categories, straight wires or coiled wires, where coiled wires have higher contraction compared to their straight counterparts [6]. Coiled SMA wire can contract by lengths of approximately 46% in comparison to their original length [6]. Although coiled SMA wires have the advantage of having higher rates of contraction, they have the drawback of having a much lower pulling force than straight SMA wires [3]. Therefore, it is desirable to use straight SMA wires when a higher pulling force is needed.
In order to preserve the benefit of having higher pulling force in straight SMA wires while maximizing the total contraction length, the straight SMA wire must be made as long as possible, since the total contraction of an SMA wire increases as the total length of the wire increases. There have been various methods of increasing the total wire length, while still arranging the wires to fit into a small space. Ballester et. al presented a method of arranging SMA wires using pulleys [2]. Copaci et. al used a method of arranging SMA wires using a Bowden cable [4]. These methods of arranging SMA wires do indeed solve the problem of a long strand of SMA wire taking up too much space, but utilizing these methods of arrangement causes friction between the SMA wires, or in case of the Bowden cable, unnecessary heat accumulation, which can cause losses in both actuation time and tension [4, 10].
Methods for Actuating Shape Memory Alloys
The first approach we took for actuating the SMA wire, proposed by former OSX intern Temma Suzuki, was to heat it up using Joule heating, which involved passing an electric current through the wire. The wire would then be cooled down using natural air convection. This method of actuating SMA wires has been widely researched, and it has the advantage of being extremely lightweight and compact. However, Joule heating presents some issues when trying to actuate the SMA wire at a high frequency. Firstly, the rate of heat dissipation in natural air convection is not as efficient when compared to other fluids (i.e., water), which slows down the cooling time [3]. The rate of cooling of the SMA wire is also dependent on the temperature of the surrounding air, which makes it difficult to cool the wire at a consistent rate [3]. Another issue with using Joule heating was the tremendous amount of electric current required to actuate the SMA wires. For instance, a single 0.1 mm diameter SMA wire requires 300 mA of current to actuate [5].
The second approach we took was to actuate the SMA using temperature-controlled water. Park et. al showed in their research that actuating SMA wires using temperature-controlled water yields a maximum actuation frequency of 1Hz [3]. This method of actuation also has the advantage of far more efficient cooling of the SMA wire compared to natural convection with air, since water has a significantly lower heat capacity than air. This method also eliminates the need to run a large current through the SMA wire.
Design Approach, and (Incomplete) Assembly
The primary goals for the design of the SMA actuator were to match the hardware specifications of the MIT Humanoid robot from the Biomimetic Lab at MIT [7], with the ultimate goal being making an SMA-actuated humanoid robot that can jump. For the SMA actuator to meet the specifications of the MIT Humanoid robot, the following hardware elements were held paramount: high force, high actuation frequency, fast speed of contraction, compact size, and low weight. To make the SMA actuator as compact as possible, the SMA wires must be fixed as densely as possible. We opted to use the 100μm diameter NT-TTR wire from FURUKAWA TECHNO MATERIAL CO., LTD [5]. According to experiments conducted on an SMA actuation testbed that employs Joule heating, developed by former OSX intern Temma Suzuki, the NT-TTR SMA wire has a maximum tension force of around 140 grams and a maximum contraction speed of 16 mm/s (for a 14 cm long wire). We first made the goal of using 1000 of these SMA wires in our actuator, which would equate to a pull force of approximately 1373 N.
Since actuating 1000 SMA wires using electricity would require 300 A of current to activate all the wires, we decided to use temperature-controlled water to control the SMA actuator, like Park et al’s work [3]. For a water temperature controlled SMA actuator to be used on a robot, there needs to be a water circulating device that can supply temperature-controlled water to the SMA actuators. To simulate this, a water circulation system was designed using an Arduino microcontroller, solenoid valves, and water pumps. The water circulation system was inspired by a water temperature controller from a thermal display in Gallo et. al’s work [8]. A hot water kettle and an ice bucket was also prepared to provide hot and cold water required for SMA actuation. To mount the SMA wires, the following linear-guide apparatus was designed.
The SMA wires are mounted to a top plate with holes placed very densely. The SMA wires are then subsequently glued into the hole, thus fixing the wires.
The apparatus is then wrapped in a large silicone tube, sealing the fittings on both ends, and thus, allowing water to flow through. The slider is then connected to a bias spring, which supplies the SMA wires with enough load stress to undergo contraction.
Assembling the Testbed
Since the internship period was not very long, the testbed was unfortunately not fully assembled at the end. The following is a sneak peak of the assembly process.
The assembly of the SMA actuator involved a process of gluing each SMA wire onto the SMA holding plate. The individual SMA wires were first passed through an arbitrarily selected hole on one of the top plates, then glued in place. Subsequently, the other side of the wire was then glued into the corresponding hole on the top plate of the opposing side. This was an incredibly tedious process.
After the individual SMA wires were glued in place, a silicone tube to seal the SMA wires from water was made using a thin sheet of silicone, 0.5 mm in thickness. The thin sheet of silicone was initially cut to size using a knife, rolled into a hose shape, then glued together using silicone sealant, and left to dry for around 30 minutes. The glued silicone sheet was then fixed to the SMA actuator using hose bands.
The testbed was planned to be controlled by solenoid valves. The diagram of the flow of water is shown below.
An Arduino microcontroller was used to control the flow of water. The circuit diagram is shown below.
In the end, we conducted an experimental run of the testbed. We discovered that the hose band was not sufficient at securing the silicone tube, which resulted in water leakage from a minor gap between the fitting and the silicone sheet held in place by the hose band. This marked the end of the internship.
Conclusions and Future Works
Due to time constraints that caused the project to be unfinished, this project is still a work in progress. Further research is needed for coming up with a way to implement this type of actuation system in a humanoid robot. For future works, the following things can be considered:
To utilize a temperature-controlled water SMA actuator in a robot, there must be a system to supply the temperature-controlled water to the SMA in a timely manner. Gallo et. al developed a water temperature control system, where water from a hot water tank and a cold water tank were mixed to output water of a desired temperature [8]. In their water circulation system, the temperature in each tank was controlled using a Peltier element and a set of heat exchangers [8].
An automatic temperature controlling water circulation device, similar to the one used in Gallo et. al’s work can be implemented into a robot. This circulation system would supply the temperature-controlled water to the SMA actuator for robot movement. For a water temperature control system like the one listed above to be implemented in a legged robot, total weight and size must be considered in the design to keep the actuator from hindering the robot’s dynamical performance. Ideally, the temperature-controlled water SMA actuation system would be as lightweight and compact as possible.
A pump that can generate just the right amount of water pressure must also be found, while maintaining a compact and lightweight formfactor. The pump that was purchased for the prototype SMA actuator testbed (EZO-PMP from Atlas Scientific) did not create adequate pressure to provide enough flow of water.
My Thoughts About the Internship OMRON SINIC X
Being an intern at OMRON SINIC X provided me the chance to live and work in Japan for an extended period of time, which also happened to be my first experience living in Japan for a long duration. Through the internship program, I was able to get a glimpse of what it is like to work in Japan. The internship also allowed me to gain real-world engineering experience and exposure to soft robotics research, which gave me the opportunity to apply what I have learned in my mechanical engineering curriculum from a school-setting to the real-world.
Throughout the internship, Dr. Tanaka and Dr. Hamaya provided great encouragement, which has further inspired me to pursue a career in robotics research. Not only do they help me cultivate my creativity and innovation skills, but they also give me the chance to foster my collaborative skills, which are extremely valuable in research.
This internship allowed me to challenge myself, and it especially taught me how to thrive in a research environment. OSX’s mission and vision of developing the near future to accommodate the needs of society has reminded me about the overarching meaning of robotics research: that is, using robotics to create a better future for society.
Again, I am truly grateful for Dr. Kazutoshi Tanaka and Dr. Masashi Hamaya for their kindness and support. I am also grateful for the researchers of OSX, as well as the other fellow interns that were a part of OSX during my time there, for their kindness and support. I am looking forward to working with them again, if I ever have the privilege of doing so. I would also like to sincerely appreciate the GA Office Center at OSX for being very supportive during my time at OSX and helping me settle down quickly to focus on my internship in Japan.
References
[1] Arias Guadalupe, Janeth, Dorin Copaci, David Serrano del Cerro, Luis Moreno, and Dolores Blanco. 2021. “Efficiency Analysis of SMA-Based Actuators: Possibilities of Configuration According to the Application” Actuators 10, no. 3: 63. Available: https://doi.org/10.3390/act10030063
[2] Ballester, Carmen, Dorin Copaci, Janeth Arias, Luis Moreno, and Dolores Blanco. 2023. “Hoist-Based Shape Memory Alloy Actuator with Multiple Wires for High-Displacement Applications” Actuators 12, no. 4: 159. Available: https://doi.org/10.3390/act12040159
[3] C. H. Park, K. J. Choi and Y. S. Son, “Shape Memory Alloy-Based Spring Bundle Actuator Controlled by Water Temperature,” in IEEE/ASME Transactions on Mechatronics, vol. 24, no. 4, pp. 1798–1807, Aug. 2019, doi: 10.1109/TMECH.2019.2928881.
[4] Copaci, Dorin, Dolores Blanco, and Luis E. Moreno. 2019. “Flexible Shape-Memory Alloy-Based Actuator: Mechanical Design Optimization According to Application” Actuators 8, no. 3: 63. Available: https://doi.org/10.3390/act8030063
[5] “Electrical Current Actuator”, FURUKAWA TECHNO MATERIAL CO. LTD, https://www.furukawa-ftm.com/tokusyu/english/product/%e9%80%9a%e9%9b%bb%e3%82%a2%e3%82%af%e3%83%81%e3%83%a5%e3%82%a8%e3%83%bc%e3%82%bf/
[6] FLEXINOL® Actuator Spring Technical and Design Data, DYNALLOY inc., Available: https://www.dynalloy.com/tech_data_springs.php
[7] M. Chignoli, D. Kim, E. Stanger-Jones and S. Kim, “The MIT Humanoid Robot: Design, Motion Planning, and Control For Acrobatic Behaviors,” 2020 IEEE-RAS 20th International Conference on Humanoid Robots (Humanoids), Munich, Germany, 2021, pp. 1–8, doi: 10.1109/HUMANOIDS47582.2021.9555782.
[8] S. Gallo, L. Cucu, N. Thevenaz, A. Sengül and H. Bleuler, “Design and control of a novel thermo-tactile multimodal display,” 2014 IEEE Haptics Symposium (HAPTICS), Houston, TX, USA, 2014, pp. 75–81, doi: 10.1109/HAPTICS.2014.6775436.
[9] “Shape Memory Alloys”, Shape Memory and Structures (SMS) Research Group, https://publish.illinois.edu/shape-memory-structures/shape-memory-alloys/
[10] Skyentific, “How bad is Bowden cables for robotics: friction tests (part 2)”, 2021, Online Video: https://www.youtube.com/watch?v=usPjM6weo8o
