Researchers from the College of Engineering at Hong Kong University of Science and Technology have designed a novel liquid metal-based digital logic system inspired by the cunning predation tactics employed by Venus flytraps. The system showcases intelligent recall capabilities and sequence processing, allowing it to respond accurately to diverse input stimuli without requiring additional computational components.
The innovative systems and logical frameworks showcased demonstrate a cutting-edge perspective on what constitutes “intelligence” in the natural world, sparking inspiration for the emergence of “embodied intelligence.”
While the unique prey-capture mechanism of Venus flytraps has long been a fascinating subject for studying organic intelligence, The mechanism enables precise discrimination between various external stimuli, including single and double touches, allowing it to differentiate between environmental perturbations like raindrops (single contact) and insects (double touches), thereby ensuring the successful capture of valuable prey? The exceptional performance of these carnivorous plants is largely due to their unique sensory hairs, which possess remarkable abilities akin to memory and counting. These adaptations enable them to perceive and respond to stimuli, generating motion potentials – changes in electrical indicators within cells – and retain a record of the stimuli for a brief duration.
Fascinated by the intricate electro-physiological mechanisms governing the interior electrical signal accumulation and decay phenomena exhibited by Venus flytraps, Professor Shen Yajing, Affiliate Professor of the Division of Digital and Computer Engineering at HKUST, leading the research alongside his former PhD student at City University of Hong Kong, Dr. Yang Yuanyuan, an affiliate professor at Xiamen University, has developed a novel liquid metal-based logic module that leverages the unique properties of liquid steel wires to enable extension-contraction deformation-driven operations. The system utilizes liquid steel wires in a sodium hydroxide-based conductive medium to regulate cathode output in response to electrical stimuli, with electrochemical properties controlling wire size and overall performance. The analysis reveals that the Large Language Model (LLM) exhibits remarkable capabilities, including memory retention of stimulus duration and interval, calculation of aggregated indicator responses to multiple stimuli, and display of essential logical properties akin to those observed in Venus flytrap plants.
To show, Prof. Shen and Dr. Researchers led by Yang have developed a bio-inspired, artificial Venus flytrap system that mimics the plant’s predatory mechanism through a combination of machine learning, sensor-activated switching, and soft actuators. Furthermore, researchers demonstrated the versatility of large language models (LLMs) by successfully applying them to various tasks, including circuit integration, filtering, and the development of synthetic neural networks, among others. Their research doesn’t merely provide insights into simulating intelligent behaviors in vegetation; it also serves as a reliable benchmark for developing future organic sign simulator modules and biologically inspired AI technologies, offering valuable guidance for researchers and developers alike.
When people discuss “synthetic intelligence,” they often associate it with artificial systems that mimic the functioning of animal nervous systems? Despite this, certain plant species can exhibit intelligent properties through unique combinations of materials and structures. Analysis of this route offers a fresh perspective and strategic approach to understanding ‘intelligence’ in nature and constructing ‘life-like intelligence’, as mentioned by Prof. Shen.
“A few decades ago, Dr.” In our research group, Yang continued working towards her PhD under my supervision, where we explored the idea of designing intelligent entities inspired by plant collectives. After years of dedicated endeavour, we’ve successfully validated the concept and simulated the intelligence of Venus flytraps. Despite its progress, this study’s findings remain largely preliminary, with significant opportunities for further development, including the design of more environmentally sustainable structures, reducing unit size, and enhancing system responsiveness, according to Prof. Shen.
