A Robot That Swims Like a Manta Ray

Engineering student Gabrielle Franzini mimics nature to make an efficient underwater device

“I found this cool thing,” Gabrielle Franzini recalled her high-school friend saying enthusiastically. “Let’s make something with it!”

The cool thing was an Arduino microcontroller and the something they made was a robot that earned them third place at their state science fair. “It was made from cardboard and balsa wood, with an ultrasonic sensor. It could basically move forward and turn left,” Franzini explained. “But I thought it was awesome and it made me realize that engineering is what I want to do.”

Figure 1.  The goal - a manta ray needs only single flap of its fins to propel itself several body lengths. Picture from SoftRobotics Toolkit.

Figure 1. The goal – a manta ray needs only single flap of its fins to propel itself several body lengths. Picture from SoftRobotics Toolkit.

Energized by her high school experiences with robotics, Franzini enrolled at Worcester Polytechnic Institute, where she has continued her winning ways. In the final year of her bachelor’s degree in robotics engineering, Franzini worked with a team of fellow engineering students to design a robotic manta ray in SOLIDWORKS that earned them first prize in the 2016 Soft Robotics Design Competition.

Soft Robotics deals with flexible structures and actuators, using materials such as silicone, rubber, or fabric instead of the aluminum or steel construction of typical “hard” robots.

“We normally work on rigid systems,” Franzini explained, “but for this project we wanted to do an underwater robot and explore efficient, low-power propulsion.” Since the manta ray is an efficient swimmer, capable of traveling long distances with minimal power output, her team felt that it would be an ideal model for their design.

“There is a good correlation between soft robotics and living organisms,” Franzini added, explaining how team was able to design the wings of its manta ray robot to achieve a complex “bio-mimetic” flexing motion using a pump, valves and a simple, low-power MSP 432 microcontroller. “The valves only drew milliamps of current,” Franzini said. “The robot was designed to glide.”

This flexibility came at a price, however. Soft robots are notoriously difficult to model due to their highly deformable nature; and it was here that Franzini’s SOLIDWORKS experience gave the team a competitive edge. As the CAD expert on the team, Franzini studied the anatomical ratios of the manta ray and created a model of the desired shape of the wing.

Figure 2.  An early, small-scale prototype showing the SOLIDWORKS simulation compared to the silicone model of the wing. Soft robots can be notoriously difficult to model, but Franzini was pleased with the simulation results she was able to achieve. (Image courtesy of Gabrielle Franzini.)

Figure 2. An early, small-scale prototype showing the SOLIDWORKS simulation compared to the silicone model of the wing. Soft robots can be notoriously difficult to model, but Franzini was pleased with the simulation results she was able to achieve. (Image courtesy of Gabrielle Franzini.)

After experimenting with various means of articulating the wing, the team chose to use a system based on the PneuNets actuator. PneuNets is an abbreviation of “pneumatic networks” and the actuators are based upon a series of channels and chambers inside an elastomeric structure. Franzini modeled the system in SOLIDWORKS and ran simulations to predict the bend in a small-scale model of their wing, as shown in Figure 2.

The team used the simulations to design the necessary arrangement of chambers to build and test a six-inch model out of silicone. “We were able to predict the bending in SOLIDWORKS,” Franzini explained. “The simulations gave us the measurements for our silicone model.”

The process Franzini used is similar to this Mold Design Tutorial on the Soft Robotics Toolkit website, while there are a number of relevant simulation tutorials on the SOLIDWORKS website.

While the team was able to use a Makerbot 2 to print the molds for its small-scale model, the full-scale, two-foot-long wing was too large for the 3D printers available to the team. It had to scale up the manufacturing process. Franzini modeled the molding setup as shown in Figure 3 to allow the team to cast the wing as two halves.

Figure 3. Franzini used SOLIDWORKS to render the mold design for her team’s award-winning biomimetic manta ray robot. Rather than using traditional hard actuators, the team used an efficient combination of pumps and pressure chambers to achieve complex motion. (Image courtesy of Gabrielle Franzini.)

Figure 3. Franzini used SOLIDWORKS to render the mold design for her team’s award-winning biomimetic manta ray robot. Rather than using traditional hard actuators, the team used an efficient combination of pumps and pressure chambers to achieve complex motion. (Image courtesy of Gabrielle Franzini.)

The wings would join to a more rigid body, which contained the power supply and actuators as shown in Figure 4. While the team had originally planned to machine the wing mold using an ABB robotic arm, a number of factors—including time constraints—led to the adoption of a more “hands-on” approach using a hot wire cutter to create a foam positive for constructing the mold.

Figure 4. Franzini’s model of the final design of the manta ray robot. It was designed to glide long distances underwater using very low energy consumption. (Image courtesy of Gabrielle Franzini.)

Figure 4. Franzini’s model of the final design of the manta ray robot. It was designed to glide long distances underwater using very low energy consumption. (Image courtesy of Gabrielle Franzini.)

Figure5 shows the final, full-scale silicone wings mounted on an acrylic control box.

Figure 5. Franzini’s team of WPI students tested the “wings” on dry land. They achieved complex wing bending motions using a low-power combination of a pump, valves and an MSP 540 microcontroller. (Image courtesy of Gabrielle Franzini.)

Figure 5. Franzini’s team of WPI students tested the “wings” on dry land. They achieved complex wing bending motions using a low-power combination of a pump, valves and an MSP 540 microcontroller. (Image courtesy of Gabrielle Franzini.)

The goal was to achieve a wing flex of 35 degrees, which would be suitable for normal propulsion, shown in Figure 5.Testing of the model demonstrated that the team had greatly exceeded this goal. The wings flexed up to 60 degrees, as shown in Figure 6.

Figure 6. Wing flex in normal operation is approximately 35 degrees. Flexing is caused by water flowing into elastomeric chambers in the upper and lower half of the wing. Water was a natural choice of hydraulic fluid for an environmentally friendly robotoperating in an aquatic environment. (Image courtesy of Gabrielle Franzini.)

Figure 6. Wing flex in normal operation is approximately 35 degrees. Flexing is caused by water flowing into elastomeric chambers in the upper and lower half of the wing. Water was a natural choice of hydraulic fluid for an environmentally friendly robotoperating in an aquatic environment. (Image courtesy of Gabrielle Franzini.)

Figure 7. The WPI Soft Robotics team’s manta ray wing at full extension. (Image courtesy of Gabrielle Franzini.)

Figure 7. The WPI Soft Robotics team’s manta ray wing at full extension. (Image courtesy of Gabrielle Franzini.)

“There were definitely challenges along the way,” Franzini said, describing how everything from valve selection to waterproofing presented unanticipated problems. “There were many late nights!”

In the end, however, graduation deadlines became one of the most pressing challenges. “We almost got it in the water… we had the fluid system going and the wings flexing,” Franzini said with perhaps a slight twinge of regret in her voice when asked about the time constraints.

Looking at the overall project, however, she has nothing but passion and enthusiasm, “We designed and built an entire robot from the ground up, with plumbing, pumps, controls… and so many iterations of the prototypes.” The team, shown in Figure 8, wrote up the results and was recognized with first prize in the 2016 Soft Robotics Competition.

Figure 8. The award-winning WPI Soft Robotics team. Left to right: Nathan Schmidt, Joshua Fuller, Gabrielle Franzini and John Price. (Image courtesy of Gabrielle Franzini.)

Figure 8. The award-winning WPI Soft Robotics team. Left to right: Nathan Schmidt, Joshua Fuller, Gabrielle Franzini and John Price. (Image courtesy of Gabrielle Franzini.)

Franzini continues to pursue her passion for CAD, robotics and competition. She is currently working towards an accelerated masters degree in robotics engineering at WPI and tutors undergraduate students on SOLIDWORKS.

She also works as an intern for SOLIDWORKS, where she is currently involved in modeling components for the 2017 FIRST Robotics Competition (FRC). FRC is one of the largest, most dynamic robotics competitions on the planet, requiring teams of high school students and their adult mentors to create large, complex machines in a matter of weeks, then bring them together for a sporting challenge of teamwork and technical challenge.

If a robot made of cardboard and balsa wood could inspire Franzini to pursue a career in engineering, then her work can be seen as a way of giving back and inspiring the next generation of students to pursue a career in engineering and design. The spirit of “I’ve found this cool thing, let’s make something,” is being passed on!

To learn more about SOLIDWORKS education programs, follow this link. If you are a researcher looking for access to SOLIDWORKS, click here.

SOLIDWORKS has sponsored this post. It has provided no editorial input. For more information, go to www.solidworks.com