Human vision is remarkably efficient. Our eyes can quickly detect movement, distinguish important objects from their surroundings, and adjust their responses depending on what we are looking at. Instead of processing every visual detail equally, the biological visual system prioritizes meaningful changes while suppressing unnecessary background information. This ability helps the brain process complex scenes without wasting excessive energy.
Now, scientists are working to bring similar capabilities to artificial vision systems. A research team led by Chao Zhao has developed an innovative organic photosensor that can detect changes in light and adjust the strength of its electrical response using a programmable voltage. By combining these functions in a single active layer, the technology could help make future cameras, robots, and artificial intelligence systems faster and more energy-efficient.
The new device, known as an ionic–electronic event-driven sensor (IEES), combines two important functions: detecting changes in a visual scene and controlling how strongly those changes are represented.
Why Conventional Artificial Vision Systems Have Limitations
Most conventional cameras capture images by measuring the amount of light falling on individual pixels. These images are then processed by electronic circuits or computer systems to identify movement, recognize objects, and understand the surrounding environment.
However, continuously capturing and processing complete images can consume considerable computing power, especially when a scene contains many details that are not important to the task.
Event-based vision sensors offer an alternative. Instead of recording complete images at regular intervals, they respond to changes in brightness. For example, when a moving object crosses a camera's field of view, the sensor can generate electrical signals corresponding to changes in illumination.
This approach can reduce redundant information and help systems respond more quickly to movement.
Nevertheless, existing event-driven sensors have limitations. Many rely on complex transistor-based pixel structures or additional analogue circuits to control their responses. Other technologies generate temporary electrical signals when illumination changes, but offer limited control over the strength of individual events directly within each pixel.
Some organic devices can also adapt to changing illumination through the movement of ions. However, their responses often depend on previous lighting conditions rather than providing independently programmable control over signal strength.
The research team sought to overcome these limitations by combining event detection and adjustable response strength within a single material structure.
How the New Sensor Works
The key to the new technology is a material system that allows both electrons and ions to move. These materials are known as organic mixed ionic–electronic conductors, or OMIECs.
Unlike conventional semiconductor materials, which primarily rely on electronic charge transport, OMIECs can support interactions between electronic charges and mobile ions. This combination allows electrical responses to be influenced by both rapid electron movement and slower ionic processes.
The researchers used two organic semiconductor materials, PgBDT-T and Y6, to create a structure known as a bulk heterojunction. This arrangement brings the materials together within the same active layer, allowing them to perform complementary roles when exposed to changing light.
The sensor generates a temporary electrical spike when the illumination changes. This spike represents a visual event, such as a change in brightness caused by movement.
Two processes work together to produce and regulate this response.
First, when the illumination changes, light generates electrical charge carriers within the active material. Electrons are rapidly extracted through the Y6 component, producing an excitatory photocurrent spike. In simple terms, the sensor briefly produces a strong electrical signal when it detects a change in light.
Second, holes, which are positively charged electronic carriers, accumulate in the PgBDT-T component. Mobile ions compensate for these charges, changing the internal electrical conditions of the material.
This process screens the electric field responsible for extracting electrons and encourages charge recombination. As a result, the photocurrent is suppressed after the initial response.
Together, these processes allow the sensor to produce a rapid signal followed by an inhibitory effect that reduces subsequent electrical activity.
This is particularly useful because the sensor does not simply measure how much light is present. It responds dynamically to changes in illumination while controlling the strength of its electrical output.
Voltage Control Mimics an Important Feature of Human Vision
One of the most interesting features of the IEES is its ability to adjust the amplitude of its electrical spikes using an applied voltage.
In biological vision, the nervous system does not treat every signal equally. Attention and surrounding context can influence how strongly sensory information is processed. Important signals can receive greater emphasis, while less relevant information is suppressed.
The new sensor reproduces a simplified version of this principle through electrical control.
By adjusting the applied bias voltage, researchers can change the strength of the sensor's response to illumination changes. This allows different visual events to receive different levels of emphasis before their information is read by an external electronic system.
For example, a vision system designed to detect moving objects could assign stronger responses to changes that are particularly useful for its task. Other events could receive weaker responses, helping reduce the amount of information that needs further processing.
This approach is called gain modulation. In the new device, the modulation is achieved through the interaction between ions and electrons rather than requiring a separate circuit to perform the same function.
Importantly, the sensor combines temporal-contrast detection and amplitude control within the same active layer. This integration could simplify the design of future neuromorphic vision systems.
Testing the Sensor With an Artificial Neural Network
To demonstrate the practical potential of their technology, the researchers incorporated the sensor into a hybrid amplitude–temporal-coded neural network, called the IEES–HATC NN.
In this system, changes in illumination are converted into analogue electrical events. Their amplitudes can be adjusted through voltage control, while their timing carries additional information about changes in the visual scene.
This combination allows the system to represent visual information using both the strength and timing of electrical signals.
The approach is inspired by the way biological nervous systems communicate through electrical activity. Rather than relying exclusively on conventional image data, the network receives signals that already contain information about changes in the scene.
The researchers evaluated the system using the Weizmann human-action dataset, which contains visual examples of human activities. According to the reported results, the approach achieved 100% classification accuracy, with a front-end energy consumption of 0.064 millijoules (mJ) under the study's matched benchmarking conditions.
These results demonstrate the potential of combining event-driven sensing with adjustable signal strength for visual recognition tasks.
However, the reported accuracy applies to the researchers' particular experimental setup and evaluation conditions. It does not establish that the sensor will achieve perfect recognition across all real-world environments or more complicated visual tasks.
Similarly, the reported energy figure describes the front-end under the stated benchmarking conditions, rather than the total energy consumption of every component in a complete artificial vision system.
What Could This Technology Be Used For?
If further developed, the sensor could contribute to several emerging technologies.
Robotics: Robots often need to detect moving objects, avoid obstacles, and react quickly to changing surroundings. Event-driven sensors could help them focus on meaningful visual changes without repeatedly processing entire images.
Autonomous systems: Vehicles and other autonomous machines must interpret dynamic scenes while operating within strict power and processing limits. Sensors that reduce redundant visual information could eventually support more efficient perception systems.
Wearable and edge AI devices: Small electronic devices have limited battery capacity. Integrating visual signal processing directly into sensors could reduce the amount of information transferred to external processors, potentially improving energy efficiency.
Neuromorphic computing: This field aims to develop computing hardware inspired by the information-processing principles of biological nervous systems. By combining excitation, inhibition, and programmable signal strength, the IEES provides a possible building block for such systems.
These applications remain potential directions for future development. Practical deployment will depend on additional work involving device reliability, manufacturing, operating conditions, and integration with complete vision systems.
A Step Toward Smarter, More Efficient Artificial Vision
The development of the ionic–electronic event-driven sensor represents a different approach to designing artificial vision hardware. Instead of relying entirely on external circuits and software to determine which visual changes deserve attention, the device uses the physical behaviour of its materials to perform part of that work directly at the sensing stage.
Its combination of rapid electron extraction and ion-mediated inhibition enables temporary electrical spikes whose amplitudes can be adjusted through voltage control. This provides a way to detect changes in light and regulate their importance within a single active layer.
Although the technology is still at the research stage, it highlights how materials science could help artificial sensors become more selective and energy-efficient.
The long-term goal is not simply to build cameras that capture images, but to develop visual systems that process information more intelligently from the moment light reaches the sensor.
By bringing event detection and programmable signal modulation together, this research offers a promising step toward artificial vision systems that borrow some of the efficiency and adaptability of the human eye.
Reference: Zhao, C., Su, X., Chen, X. et al. Amplitude-controllable event-driven organic photosensors based on ionic-mediated inhibition. Nat. Mater. (2026). https://doi.org/10.1038/s41563-026-02747-8
Comments
Post a Comment