Haifeng Ling

Micro‑unmanned aerial vehicles urgently require compact, lightweight, and low‑power conformal vision modules capable of wide‑field imaging, polarization perception, and illumination‑adaptive processing. However, the monolithic integration of intrinsic stretchability, polarization sensitivity, and neuromorphic adaptation within a single deformable optoelectronic platform has remained a major challenge. Here, we report an intrinsically stretchable neuromorphic vision transistor array based on a semiconducting composite film consisting of an organic conjugated polymer (TDPP‑Se), CdSe/ZnS quantum dots (QDs), and an elastomeric SEBS matrix. The QDs significantly enhance photoresponsivity and enable gate‑tunable light adaptation, while shear‑aligned polymer chains yield strong polarization sensitivity with a high polarization ratio. The devices integrate sensing, memory, and in‑sensor computing with an ultralow energy consumption of ∼0.56 pJ per synaptic event, and retain stable multimodal performance under tensile strains up to 50%. When conformally laminated onto hemispherical substrates, the system achieves a mechanically tunable FOV from 53° to 149°. When the experimentally measured array responses are computationally mapped onto image preprocessing, the resulting vision framework achieves a classification accuracy of 95.6% under dynamic illumination, strong background interference, and large mechanical deformation. This work establishes a conformal neuromorphic vision platform for robust and energy‐efficient artificial vision in micro‐autonomous systems.

In the field of ultrafast optoelectronic memory technologies, organic field-effect transistor (OFET) memory has attracted considerable attention due to its multifunctionality, tunability, and low power consumption. Among its components, the charge-trapping layer plays a critical role in nonvolatile OFET memory design. Proper modulation of this layer can enhance memory performance. To further enhance the programming speed and stability of the device, it is essential to construct large-scale multifunctional nanostructures with tunable nanostructure dimensions within the charge-trapping layer. Herein, blending poly(9-vinylcarbazole) and poly(9,9-dioctylfluorene-2,7-diyl) with distinct chain lengths induces phase separation due to their surface energy difference, enabling the formation of a densely distributed, large-scale nanopillar array with tunable dimensions. A conformal interface with a semiconductor layer is subsequently achieved. This precisely controlled interface enhances trap-assisted tunneling, accelerating charge injection and facilitating the Fowler–Nordheim tunneling process. The devices demonstrate ultrafast nonvolatile memory performance, including a 50 μs programming speed and sustained retention exceeding 22 000 s. Architectural periodicity in patterned polymeric nanostructures has been demonstrated to govern charge-trapping efficacy. Conductive atomic force microscopy reveals the operational mechanisms by which the ordered interfaces fundamentally regulate the carrier injection dynamics, enabling the realization of organic functional integrated circuits with ultrafast nonvolatile memory matrices.

Biological neuromuscular systems seamlessly integrate signal sensing, synaptic processing, and motor execution, all while possessing remarkable self‐healing capabilities to maintain functional stability after injury. Mimicking this sophisticated integration of neural signal processing and action with embodied resilience remains a grand challenge for artificial systems. Here, we report a self‐healing ionic synaptic device that leverages ion‐mediated signal transmission and dynamically reversible borate ester bonds within a bilayer hydrogel matrix. The device emulates key synaptic functions, including short‐term plasticity, paired‐pulse facilitation, and experience‐accelerated learning. More importantly, it exhibits bio‐inspired dynamic recovery capabilities, autonomously restoring its synaptic functions after severe mechanical damage (100% recovery in Young's modulus). By integrating this device into a neuromorphic control circuit, we construct an artificial neuromuscular system that successfully translates pulse signals into precise, graded movements of a mechanical arm, mimicking muscle contraction and relaxation. The system maintains robust performance even after device damage and self‐healing, demonstrating features such as Morse code recognition and Pavlovian associative learning. This work provides a paradigm for developing resilient, brain‐inspired intelligent systems with perception‐action capabilities for soft robotics and adaptive human‐machine interfaces.

In‐sensor vision computing system as an emerging edge computing platform shows great potential to process dynamic information. However, it still requires an electric to reset weight, largely limiting its capability in processing complex signals. Here we propose FeO _x optomemory that can automatically convert its states from negative photoconductance memory (NPM) to positive photoconductance memory (PPM). The formation of neutral oxygen vacancy ( V _o ) from both the photogenerated electron‐based iron reduction process (Fe ³⁺ to Fe ²⁺ ) and the photoelectron trapping by charged oxygen vacancy ( V _o ^x+ ) builds the NPM effect under low light dosage illumination. The decrease in V _o by the photogenerated hole‐based iron oxidization process (Fe ²⁺ to Fe ³⁺ ) and the increase in V _o ^x+ by Joule‐heating assisted photoelectron detrapping from V _o sites causes the automatic conversion from the NPM to the PPM when the light illumination exceeds the threshold dosage (0.72 µJ/µm ² ). Such self‐adaptive conversion from NPM and PPM enables the FeO _x optomemory to execute fully optical computing. The NPM effect provides rich echo states for dynamic feature encoding while the PPM initializes the encoded states, thus building a self‐adaptive reservoir computing (RC) system, yielding a recognition accuracy of 97.93%. This work provides an emerging in‐material encoding mechanism for in‐sensor edge computing system.

Organic neuromorphic visual sensors promise to overcome the limitations of conventional image sensors, but a trade-off between efficient charge transport and charge trapping has hindered the development of interfaces that combine efficient optical sensing and non-volatile memory. Here we show that an inorganic-organic heterogeneous dielectric (Al ₂ O ₃ /PAA), fabricated by low-temperature plasma-enhanced atomic layer deposition and spin-coating, enables flexible thin-film transistors with integrated sensing-memory-processing functions. The devices exhibit an average mobility of 22.65 cm ² V ⁻¹ s ⁻¹ , a 30 µs optical response to 450 nm light, charge retention exceeding ten years, write/erase endurance over 10 ⁴ cycles, an electrical response down to 130 ns, and stable multilevel programming at 10 µs. They emulate synaptic plasticity and achieve recognition accuracies of 99.49% (gesture), 94.78% (digit) and 92.56% (face). When integrated with a convolutional neural network accelerator, the system demonstrates real-time, interference-resistant face detection, highlighting the potential of this dielectric architecture for flexible neuromorphic vision systems.

The flexible integration of contact perception and response actuation in a single soft system without signal crosstalk remains a key challenge in robotics and human‐machine interaction. Inspired by the division of human hand function, we present a monolithic cross‐talk‐free gripper system with photo‐thermal‐driven soft fingers and a tactile‐perceiving palm. It achieves physical decoupling of electro‐tactile sensing and photo‐thermal actuation through a structurally partitioned, biomimetic design. The central sensor and surrounding actuators are fabricated via patternable screen printing on a single matrix to form a bilayered film. The constructed tactile sensor has a wide pressure range of 0–2500 kPa, a 5 ms response time, and over 40 000 cycle durability. The bilayer actuator is multi‐responsive to light, heat, and humidity, showing biomimicry and programmability. The strip actuator can bend 360° and lift 209 × its weight under infrared heat. Assisted by a 1D convolutional neural network, the gripper system can recognize objects in real‐time and then perform closed‐loop feedback grasping of specific objects under light irradiation. This work presents a facile and sustainable approach to creating intelligent, integrated soft robotic systems.

The challenge of achieving Ohmic contacts fundamentally limits the performance of organic electronic devices, primarily due to high contact resistance resulting from Schottky contact at the metal–organic semiconductor contact interface. This study demonstrates a general strategy to significantly diminish contact resistance in copolymer organic field-effect transistors (OFETs) by achieving a quasi-Ohmic contact where the dominant mechanism is tunneling injection rather than thermionic emission. By introducing a solution-processed dopant-assisted modification (DAM) layer fabricated via simple orthogonal solvent processing at the interface, a heavily doped interfacial region is created. This region dramatically narrows the Schottky barrier width, enabling efficient charge injection via the tunneling injection enhancement effect. An optimized DAM layer achieves an ultralow contact resistance of 66 Ω cm, representing an 85% reduction vs the conventional device. Concurrently, the field-effect mobility increases by 273% to 1.68 cm2 V−1 s−1, while the threshold voltage decreases by 36% to −9.37 V. The DAM layer exhibits broad applicability and good stability, while also showing potential for flexible OFETs. This study provides a simple and universal interfacial engineering pathway to high-performance organic transistors, paving the way for advanced organic integrated circuits.

Inspired by the auditory system's capacity to process spatiotemporal sound patterns, voiceprint recognition plays a vital role in identity authentication and security. However, current platforms often face challenges of speech frequency and amplitude variability, hindering accurate feature extraction in noisy environments. To address these issues, a large-scale hybrid metal-halide dynamic memristor (MHDM) featuring an engineered gradient-distributed architecture is developed for adaptive voiceprint recognition. The spontaneously graded metal-halide functional layer allows for precise modulation of Schottky barriers and redistribution of interface charges. This design achieves µs-scale response, enhances noise tolerance (over 20% improvement in signal-to-noise ratio), and enables kHz-scale dynamic signal processing. Experimental results demonstrate that the MHDM achieves a voiceprint recognition accuracy of 99.3%, maintaining high performance at 93.2% even in realistic background noise. These findings demonstrate the system's potential for secure and efficient voiceprint recognition, combining scalability with robust performance in noisy environments.

Mimicking the human eye’s ability to autonomously adapt to diverse and mixed illumination conditions remains a fundamental challenge in artificial vision systems. Although substantial progress has been made in materials and device engineering, current adaptive vision architectures still depend heavily on complex circuitry or algorithms and are typically restricted to uniform illumination owing to the strong intensity-dependence of photosensitivity. Here, this work presents a highly adaptive TiO₂/PEDOT:PSS photomemristor that leverages the tunable conductivity of PEDOT:PSS together with the optoelectronic response of TiO₂. The photothermal effect dynamically modulates the water absorption/desorption equilibrium in PEDOT:PSS, enabling reversible suppression or enhancement of photosensitivity under bright or dim illumination, respectively. By combining with artificial neural networks (ANNs), the artificial vision system based on TiO₂/PEDOT:PSS photomemristor arrays achieves a high accuracy of 91.3% in image recognition under mixed-light conditions—without the need for complex circuitry or algorithms. This work may establish a new approach for designing autonomous, efficient, and high-performance neuromorphic vision systems to advance the development of autonomous driving and humanoid robots.