From sky to sea: A biologically inspired aerial–aquatic robot

There is a long history of biologically inspired design and engineering in AI. The initial concept of the artificial neural network was roughly based on the way neurons in a brain activate in response to a stimulus. More recently, researchers have built four-legged robots that can run, jump, and climb like a dog or mountain lion. 

Ke Wu, Visiting Assistant Professor of Robotics at MBZUAI, is continuing this tradition by designing soft, bio-inspired robots that draw inspiration from evolutionary processes. Wu explores how principles from nature and evolution can inform not only the physical design and control of robots but also the development of intelligent systems that can learn and adapt in ways similar to living organisms.

“There are fascinating similarities and fundamental differences between biological and artificial learning,” Wu says. “In nature, knowledge is encoded and passed down genetically from one generation to the next, allowing species to evolve gradually over time. Similarly, intelligent systems can accumulate experience and improve their performance through learning.”

“However, unlike biological evolution, where knowledge transfer is slow and limited to descendants, robots can instantly deploy learned information or skills to any other robot, greatly accelerating collective learning and adaptation.”

Wu has worked on many bio-inspired robots, including soft continuum robots modeled after elephant trunks, snakes, and octopuses. His research extends beyond biological forms to explore how structural intelligence and movement patterns shaped by evolution can guide robot design, control, and learning. By integrating AI and the principles of embodied intelligence, Wu’s work enables robots to learn and adapt more efficiently while offering new insights for developing more adaptive, biologically grounded intelligent systems.

One of his recent projects is a transmedia drone that is designed to move in both air and water and is inspired by the diving beetle, an insect that can fly and swim.

Wu and colleagues will present their new design at this year’s IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2025) this month in Hangzhou, China.

Junzhe Hu, of Carnegie Mellon University and the University of Cincinnati, and Pengyu Chen, Tianxiang Feng, Yuxuan Wen, and Janet Dong of the University of Cincinnati, are co-authors of the study.

Flexibility as strength

Soft robotics is an emerging field that combines robotics, bionics, and materials science. With the rapid advancement of AI, the field is evolving at an unprecedented pace to meet growing demands for intelligent, adaptable, and energy efficient robotic systems. Wu describes it as an extremely active area of research in which researchers are rapidly proposing and exploring new ideas.

When people hear the word “robot,” they most likely think of what is known as a rigid robot. These are built out of hard materials and use mechanical joints to move.

Rigid robots have been used for decades in controlled and controlled environments, like warehouses and factory floors. While rigid robots are precise and powerful, their reliance on rigid mechanical joints constrains their degrees of freedom and maneuverability, resulting in limited compliance and reduced suitability for safe and intuitive human–robot interaction.

Alternatively, soft robots, with their high degrees of freedom, compliance, and adaptability, can move more freely and operate in environments that rigid robots can’t.

For instance, after an earthquake, a soft robot could search for survivors trapped under rubble by deforming to fit through narrow spaces. In medical applications, soft continuum robots can navigate within the human body with minimal invasiveness, enabling delicate procedures that would be difficult or unsafe for traditional rigid robots.

Moreover, because they are made of supple and flexible materials, soft robots can be safer in situations where they might encounter people.

Lessons from the diving beetle

Wu and his team looked to diving beetles because they can move efficiently in both air and water. These bugs use their wings to fly in the air; when they enter the water, they use hairs on their legs to help them swim.

The team’s design is based on a quadcopter drone but replaces the traditional design with what they call a variable-stiffness propulsion module. In the air, the arms of the drone are rigid. When the drone enters the water, two of the arms become flexible, allowing it to propel itself with a swimming motion.

It’s the first drone design that replaces the traditional, rigid frame of a quadcopter with a variable-stiffness propulsion module that allows it to move in both air and water.