Researchers Develop Robot That Can Grow Its Own Electronic Sensors

Researchers at the University of Tartu in Estonia and the Italian Institute of Technology have developed a robot capable of creating an electronic sensor inside its own body. The system uses a network of fluid-filled channels to deliver chemicals through a 3D-printed structure, where exposure to ultraviolet light causes a new conductive material to form.

The researchers demonstrated the technology using a moth-shaped robot with artificial wings. Before the sensor was formed, the robot couldn't respond to UV light. Once the chemical reaction had taken place, the new material became sensitive to ultraviolet radiation, allowing the robot to detect the light and activate its wings or flash an LED.

The work is described in a study titled Receptogenesis in a Vascularized Robotic Embodiment, published in Advanced Science and highlighted by the Estonian Research Council on October 30. The researchers have called the process receptogenesis, referring to the creation of new sensory receptors within an existing robotic structure.

It's an unusual approach to robotics because the researchers aren't simply adding another sensor to the machine. They're using chemical reactions to change the electrical properties of the material from which the robot is made.

How a Robot Can Grow an Electronic Sensor

The inspiration for the project came from moths and their circulatory systems. Insects have an open circulatory system that distributes a fluid called hemolymph throughout their bodies, carrying nutrients and other substances to different tissues.

Moth wings are particularly interesting because they contain networks of veins that transport this fluid. The wings are also relatively flat, making them a useful model for a robotic structure where liquid needs to be distributed over a large surface area.

The researchers used the blue underwing moth, Catocala fraxini, as their biological reference. They then created an artificial version using transparent polyethylene terephthalate glycol, or PETG, a commonly used material in 3D printing.

The robot contains a network of channels that carry liquid through its body and wings. Some are deliberately printed as larger veins, while smaller channels are formed by the gaps between the individual lines of deposited plastic.

This is an interesting part of the design. Anyone familiar with fused deposition modelling will know that the material is deposited in layers, with the internal structure determined by the printing pattern. Rather than trying to eliminate the gaps between the printed strands, the team allowed liquid to pass through them. This gave the robot a much finer network of pathways than the larger channels alone.

The robot uses a single pump to move around 3 millilitres of liquid through its body and wings. Some of that liquid also reaches into the fine spaces within the PETG, rather than remaining entirely inside the larger passages.

For the UV experiment, the liquid contained pyrrole, a light-sensitive chemical initiator and a cellulose-based additive.

Shine ultraviolet light onto the prepared area, and the pyrrole begins forming polypyrrole, a polymer that conducts electricity. The light starts the reaction, allowing the researchers to control where the new material appears.

You can see where the reaction has taken place because the new polymer forms a dark patch inside the translucent plastic.

Using a stencil to control the exposed region, the researchers were able to produce patterns with a resolution of approximately 0.3 millimetres.

In the experiments, about one minute of exposure to a 365 nm UV LED was sufficient to form the new material.

But producing something electrically conductive is only part of the achievement. The material also needs to respond to its surroundings if it is going to function as a sensor.

Polypyrrole has an interesting property in this respect. Expose it to UV light and its electrical impedance temporarily falls. That change gives the robot something it can measure, allowing the newly formed material to act as a light sensor.

The researchers had already built electrical contacts into the structure and connected them to an ATtiny43U microcontroller. Once the polypyrrole formed between those contacts, the controller could detect changes in the sensor's electrical response using its analogue-to-digital converter.

The reading could then be used to activate an LED or drive a MOSFET connected to a shape-memory alloy wire. Passing current through the wire heated it, causing it to shorten and pull the robot's wings into a different position.

The wings then returned towards their original position as the wire cooled.

This isn't a flying robot, and the wing movement is relatively simple. Nevertheless, the demonstration showed that the machine could acquire a sensing capability that wasn't present in its original material.

The sensor also continued working after more than four months of storage, although the researchers discovered that its electrical characteristics were not entirely stable.

Could Robots Eventually Manufacture Their Own Hardware?

A conventional robot leaves the workshop with its sensors already fitted. If it needs to measure something else later, somebody usually has to find room for another component, wire it in and make sure the electronics can read the signal. That's not particularly appealing if the machine is already working somewhere difficult to reach.

The researchers are investigating whether some of those changes could instead be made within the robot's existing structure.

In their demonstration, the artificial circulatory system delivered the necessary chemicals throughout the body, but a sensor only formed in the areas exposed to sufficient UV light. This could eventually allow a robot's surroundings to influence where new sensing regions develop.

There are some obvious limitations, however.

The experiment used a predetermined chemical mixture that could produce one particular type of sensor. The researchers also controlled when the chemicals entered the robot and used a stencil to determine where the sensor would form. The robot wasn't independently deciding which electronic components it needed.

There's also the question of how useful these chemically formed sensors would be compared with conventional components.

A commercial UV photodiode is small, inexpensive and has predictable electrical characteristics. Engineers can select a device with a known sensitivity, response time and operating range, then design the associated circuitry accordingly.

The polypyrrole sensors are much less predictable. According to the study, their electrical resistance varied considerably between samples, and repeated exposure to ultraviolet light caused their sensitivity to change.

This happens because some of the original pyrrole remains inside the material after the sensor has formed. Further UV exposure can continue the polymerisation process, changing the structure and electrical behaviour of the sensor.

The researchers regard this as an interesting property because the material effectively retains a physical record of its previous exposure. For conventional sensing applications, however, a sensor that changes its sensitivity during operation presents a considerable calibration problem.

There are other practical questions. A robotic system would need to carry suitable chemicals, circulate them reliably and manage the reaction conditions. The electrical connections also need to be available wherever new sensors are expected to form.

In the current demonstration, those contacts were already embedded in the structure before the new material was created. The electronics and actuator were conventional components, and the wing mechanism used external power.

So while the robot can form a new sensor material, it cannot yet manufacture an entirely new electronic circuit wherever one is needed.

Even so, I think the underlying idea deserves attention.

Robotics research has made considerable progress in giving machines more sophisticated software, allowing them to recognise objects, learn movements and adapt their behaviour. Making similar changes to the physical hardware is considerably more difficult.

This experiment approaches that problem through materials chemistry. Instead of modifying a robot by attaching new components or rearranging its existing structure, the researchers have demonstrated that its body material can be given a new electrical function after manufacture.

Whether this becomes practical outside the laboratory will depend on developing more useful chemical systems, improving the consistency of the sensors and finding applications where the ability to create a sensing region on demand offers a genuine advantage.

For now, the moth robot is a relatively simple demonstration. But the idea of an electronic device that can acquire new sensing capabilities through chemical changes within its own structure is an intriguing one, particularly when so much of modern robotics still depends on hardware that remains fixed from the moment it leaves the factory.