Can a robot think without electronics?
Sergio Picella defended his PhD thesis at the Department of Mechanical Engineering on March 4th.
What if machines could sense, decide, and act鈥攚ithout a single electronic chip?
That is the bold conclusion of Sergio Picella. His PhD research demonstrates that true mechanical autonomy is possible: by carefully designing soft materials and fluidic circuits, robots can run 鈥減rograms,鈥� harvest their own energy, generate complex behaviors, and operate unsupervised鈥攗sing only physical principles. In short, autonomy does not necessarily require electronics. It can be built directly into matter.
Human fascination with autonomous beings long predates modern robotics. Stories like Frankenstein and Pinocchio questioned whether artificial creations could think, act, or even possess a will of their own.
Today, robotics research revisits these timeless questions鈥攂ut through engineering rather than fiction.
This thesis of focuses on soft robots: flexible, compliant machines powered and controlled by fluids such as air or liquid. Unlike rigid robots with electronic processors, these systems rely on material properties, mechanical instabilities, and fluid dynamics to produce behavior. The guiding idea is simple yet powerful: structure itself can compute.
A new programming language without electronics
One of the thesis鈥� central breakthroughs is the creation of a vocabulary of electronics-free pneumatic circuits.
Traditional robots use electronic code with statements like If, 滨蹿鈥濒蝉别, and For. In this work, those logical instructions are physically embodied in networks of air channels and valves. By defining a dictionary of fluidic logic elements, the research shows how coding variables can be materialized directly in hardware, how logical decision-making can emerge from air pressure pathways, and how behavioral switching can occur automatically when the robot interacts with its environment.
In other words, the 鈥渟oftware鈥� is built into the robot鈥檚 body. Programs are not uploaded鈥攖hey are manufactured as part of the structure itself.
Energy from the environment
Autonomy requires more than decision-making; it demands energy independence.
To meet this challenge, the researcher proposes an energy-harvesting strategy that allows soft robots to operate without batteries or external power supplies. The system relies on low-boiling-point fluids and their phase transitions, combined with a fluidic control strategy that exploits naturally occurring daily temperature and sunlight cycles.
By converting environmental circadian oscillations into pneumatic power, the robot can function without electronics or human intervention. Demonstrations conducted both indoors and outdoors show that these environmental variations are sufficient not only to power the robot but also to control its behavior. Inspired by natural systems that efficiently convert energy and separate processes across different time scales, this approach opens the door to long-term, unsupervised operation.
The power of coupled oscillators
Another major contribution of the thesis lies in understanding how patterns emerge in networked systems.
Rather than centrally controlling every movement, the research investigates networks of coupled oscillators inspired by biological central pattern generators. By carefully designing how these oscillators interact, distinct and stable movement patterns can spontaneously arise.
The researcher developed analytical and numerical tools to predict such emerging behaviors and to compare different coupling strategies. It distinguishes between explicit coupling, where oscillators are directly connected, and implicit coupling, where interaction arises through the shared hardware and environment. This decentralized approach demonstrates how coordinated and adaptive behavior can emerge from simple interacting components, a principle widely observed in living systems.
Automatic vs. autonomous
While the technical achievements are substantial, the thesis also addresses a fundamental conceptual question: what does autonomy truly mean?
The literature offers no universally accepted definition, which complicates both research and interdisciplinary dialogue. To clarify this issue, the work distinguishes between automatic systems, which respond predictably to stimuli according to preprogrammed rules, and autonomous systems, which are capable of self-sustained and unsupervised operation supported by internal regulation.
By exploring the philosophical and ethical dimensions of autonomy alongside its technical aspects, the thesis proposes a clearer framework for discussing artificial agents and their role in society.
Why mechanical autonomy matters
The contributions of this research demonstrate that hardware-based programming vocabularies can be realized through fluidic circuits, that energy autonomy can be achieved through environmentally driven phase transitions, that emergent behaviors in coupled oscillators can be systematically designed and analyzed, and that a clearer conceptual distinction between automatic and autonomous systems is both necessary and possible.
Together, these advances show that autonomy can be encoded in structure, material properties, and physical interaction alone.
Beyond the lab
Soft robots with embedded mechanical intelligence may operate where electronics struggle, such as in harsh outdoor environments, radiation-heavy settings, or remote natural locations. Their compliant nature also makes them inherently safer for interaction with humans and better suited to unpredictable surroundings.
More broadly, this research suggests a shift in how we design machines. Rather than relying exclusively on increasingly complex electronics, we can engineer matter itself to behave intelligently.
Autonomy, it turns out, might begin not with silicon chips, but with air, fluid, and carefully designed materials.
Title of PhD thesis: . Supervisors: Prof. Bas Overvelde, and Dr. Erik Steur.