TECH OFFERS

Manufacturing plants constantly seek opportunities to save energy, reduce cost, and be more environmentally sustainable. However, achieving these goals often requires heavy expenditure in the form of hiring teams of experienced engineers, who then perform cost-reduction tasks manually - this method is time-consuming, costly, and prone to inaccuracies due to the risk of human error.  This technology offer provides a no-code Artificial Intelligence (AI) powered platform that monitors energy consumption, carbon dioxide(CO2) emission, and operational expenditures (OPEX) in real-time. The AI engine builds a virtual cognitive model (digital twin) of a physical asset, e.g. a manufacturing plant or a piece of machinary. Simulations are carried out on the model to predict operational inefficiency i.e. high energy usage, equipment breakdown, etc. Upon detection of inefficiencies, the engine is able to suggest the best operating parameters to resolve the inefficiency. Monitoring: Tracks real-time operational data through sensor data from every equipment Monitors the lifecycle and performance (energy usage, carbon emission, operational expenditure) Predicts and alerts to potential equipment failures Optimisation: Autonomously optimises the operational variables to prevent operational failures, reduce downtime, energy usage and carbon emission based on a user-defined thresholding value Simulation: Software comprises a simulation capabillity to test if changes in specific operating parameters can cause knock-on issues or increase efficiency The software platform can be deployed securely on-premise, private cloud, or public cloud. The technology can be paired with sensor solutions and 3D modelling software as end-to-end solutions to build digital capabilities in optimising and visualising operations/processes. This technology offer provides an AI-powered cognitive digital twin that is applicable for all types of machinary used in manufacturing operations, and refineries in the following industries: Chemical Oil and gas Pharmaceutical Energy/Power This AI-enabled solution is intended to assist in the autonomous reduction of downtime, operational expenditures, energy consumption, and CO2 emissions.   In comparison with conventional digital twin software which virtually represents physical assets with 3D models, and are commonly used as simulation, prediction, and life cycle monitoring tools. This technology can be differentiated in the following ways: Operates autonomously Does not need to be operated by specialised engineers with technical experience; workforce reduction Is not simply a complementary tool for analysis, operational oversight and decision-making Built-in AI engine acts, makes decisions autonomously to optimise throughput The technology owner is looking to collaborate with machinary owners in the chemical and process industries, as well as original equipment manufacturers (OEM) and digitisation/digital transformation companies. Cognitive Digital Twin, Optimisation, Emission Reduction, Digitilisation, Modelling, Simulation Infocomm, Artificial Intelligence, Computer Simulation & Modeling

This technology offer is technique that can turn renewable cellulose paper feedstock into lightweight carbon foams that are architected into highly complex geometries that cannot be produced through traditional manufacturing techniques, such as closed-cell lattices. These carbon foam lattices exhibited excellent mechanical properties, particularly in energy absorption, as well as good battery characteristics, low thermal conductivity and relatively good electrical conductivity. Unlike most traditional carbon foams that are brittle, paper-based carbon foams can withstand ~ 30% strain before significant deformation sets in. These multifunctional properties, the quick and easy customization of part geometry and the use of green feedstock are expected to be useful for aerospace, automobile, sports, medical and thermal insulation markets, as they search for the next generation of eco-friendly, high-performance materials. This technology is available for R&D collaboration, IP licensing, or test bedding, with partners such as battery manufacturers, supplier to battery manufacturer, space industry, etc. Technology is a 3D printed, highly porous carbon material produced from paper Density = 0.2 – 0.6 g/cc (79 – 92% porosity) Modulus = 4.5 – 700 MPa Max. Compressive Strength = 0.2 – 13.4 MPa Mechanical Energy Absorption = 0.02 – 4.8 MJ/m3; 0.08 – 10 kJ/kg Pseudo-elastic compressive strains up to 30% - 40% All the specifications above can be controlled through geometrical design As anode for Li-ion battery, specific capacity = 65 – 140 mAh/g for >300 cycles Thermal Conductivity = 0.1 – 0.5 Wm-1k-1 Electrical resistivity = 0.002 – 0.04 Wm The potential applications are as follow: Thermal Insulation Fire-proofing Refractory Material Composite Parts Rocket Nozzles Acoustic Tile EMI Shielding Water and Gas Filters Li-Battery Anodes, Aerospace Materials Material is multifunctional and covers several markets, including but not limited to aerospace, automobile, carbon foam batteries, insulation etc. The structural carbon market is ~ USD$4 billion in the US alone. The primary differentiator in this technology is that it uses cellulose paper, a freely available and renewable green resource rather than fossil fuel as a precursor. Next, the technology enables the carbon foams to be additively manufactured into prescribed geometries. Customization of geometry is therefore easy, quick and cheap and machining is not needed. Feedstock is a green, cheap, renewable material – cellulose paper Easy customization of carbon foam geometry with quick turnaround time Thermal, electrical, mechanical and battery performance are in the range of the best performing carbon foams on the market Carbon Foam, Additive Manufacturing, 3D Printing, Green Materials, Cellulose, Battery, Thermal Insulators Manufacturing, Additive Manufacturing

Existing methods for load monitoring typically focus primarily on residential building data, while few look at the effectiveness of such systems for industrial or commercial buildings. Apart from the use of this technology for real-time supply-demand response, such methods can be extended for use in anomaly detection, small-scale load change detection, or an estimation of energy usage, without the associated high costs of sub-metering equipment. The proliferation of neural networks for such demanding tasks solves the computationally expensive problem of traditional methods like Hidden Markov Models (HMM) and fuzzy clustering algorithms. This technology offer is a neural network solution for residential and industrial energy management. It utilises a time-series forecasting tool to predict load, renewable energy generation, and electricity prices, without the need for costly sub-metering equipment. It is based on reinforcement learning algorithms which are trained by rewarding and penalising neural network algorithms for good or bad decisions respectively, the solution is a non-intrusive technique that helps residential and commercial end-users save on energy costs in the open energy market by scheduling their load demand for heating, ventilation, air conditioning (HVAC) systems, washing machines, and charging of their Electric Vehicles (EVs). This technology is an integrated platform that consists of the following components: Non-intrusive load monitoring (NILM) and data analytics tools for smart homes Time series forecasting tool for renewable energy and dynamic electricity pricing Reinforcement learning-based neural network for energy management systems Electricity plan recommendation tool for residential and commercial users Support data imputation - tolerant to missing data by estimating values that are missing Integrates with several types of Deep Neural Network (DNN) models: Long Short Term Memory (LSTM) Bidirectional Short Term Memory (Bi-LSTM) Time Distributed Dense Layer The technology can be deployed for use in smart buildings, smart homes, and for commercial/industrial applications such as smart factories, server farms, etc to enable the following applications: Anomaly detection e.g. fault detection Small load change detection Energy data analytics for energy monitoring Energy disaggregation (addresses the problem of separating the electricity usage into individual disaggregated components) Determine equipment on-off status Non-intrusive load monitoring (NILM) represents a cost-efficient technology for observing power usage in buildings. It tackles several challenges in transitioning into a more effective, sustainable, and digital energy efficiency environment. Compared with existing smart home management systems that use model-based methods and only consider simple objectives, this technology helps to reduce energy costs by shifting electricity load demand to a low electricity price period while ensuring that electricity consumption needs are still met. The technology owner is interested to collaborate with smart building operators, in-home integrated system suppliers, and smart appliance manufacturers to test-bed or collaborate to build new products/services. Infocomm, Big Data, Data Analytics, Data Mining & Data Visualisation, Artificial Intelligence, Energy, Sensor, Network, Power Conversion, Power Quality & Energy Management, Sustainability, Sustainable Living, Low Carbon Economy

Machine Learning (ML) and Deep Learning (DL) have been the primary growth driver of Artificial Intelligence (AI) and has seen widespread adoption in areas such as Computer Vision, Speech Processing, Natural Language Processing, and Graph Search, among many others. It is also well-known that AI both needs and produces large amounts of data. However, traditional data repositories have not scaled effectively to handle the large amounts of vector representations that are common in AI applications - in such cases, searching for similarities across high-dimensional vectors is inefficient. To address such limitations, vector databases have been developed to address the limitations of traditional hash-based searches and search scalability, enabling similarity searches across large datasets. This technology offer is a unified Online Analytical Processing (OLAP) data platform that supports approximate vector search, enabling efficient searching over billion-scale structured data and vector data. The data engine simplifies the process of building enterprise-level AI applications such as search and recommendation systems, video analytics, text-based searches, and chatbots while accelerating the development of production-ready systems. Developers no longer need to deal with complicated scripts to query vector data as low latency, high-performance structured data, and vector data searches are made possible via vector data indexing methods and the use of extended Structured Query Language (SQL) syntax. This technology offer is purpose-built OLAP database, CPU-only implementation with a built-in vector query engine that uses extended SQL statements for data querying. Supported data include structured data (tabular text, numbers, dates, times) and unstructured data (image, video, audio) that have been converted to vector data representation. This technology enables high-performance joint queries, and a simplified manner of querying labels, text, and numbers within a single SQL statement. It supports highly performant SQL + vector searches, operating on billion-scale data, with an operating latency of 200 milliseconds at a throughput of 200 queries per second (QPS).  The key features of this technology are as follows: Fast query performance Column-oriented storage; data is stored in the same column and compression techniques are applied to reduce disk usage and save I/O resources Linear scalability Data is stored evenly across nodes, ensuring scalability Simultaneous data input Data can be inserted simultaneously via random data distribution Concurrent queries  Simultaneous insertion and querying The following similarity metrics are currently supported: Euclidean Cosine Similarity Dot Product The following indexing libraries are currently supported: Facebook AI Similarity Search (FAISS) hnswlib (with proprietary optimisations) The following interfaces are available for developer integration: C++/Java/Python language Software Development Kit (SDK) SQL interface Web User Interface (UI) This technology can be applied for similarity searches (identifying similar high-dimensionality vectors), or classification (locating images that contain a certain element, e.g. car, flower). The following potential applications of this technology have also been tested: Biometrics (fingerprint matching) DNA/genetic sequences and other biomedical fields (similarity search/classification) Multimedia - image, video, and audio (similarity search) Text-based - recommendation systems, chatbots (similarity search) Molecular (similarity search) Trademarks (similarity search) Commodities  GIS vectors (vector-based semantic analysis) Compared with existing techniques, this technology represents a single, unified pipeline for querying vector representation data without the need to store structured data and vectors separately in traditional databases (SQL) and vector repositories. This solves the limitation of having to merge results from standard database engines (specifically optimised for hash-based searches) with that of vector query databases. This data engine includes a vector search function and it can efficiently store, index, and manage vectors that are generated by deep learning networks and machine learning models. Additionally, the extended SQL query syntax of this technology enables a highly efficient, simplified search across a variety of different AI applications. The technology owner is keen to collaborate with companies that are conducting in-house AI application development in industries such as, but not limited to, e-commerce, video analytics, smart city, and healthcare. The following is an example of how the vector search engine can be used to query for similar logos (images): A large dataset of logos is prepared A feature extraction model is trained from the dataset e.g. DarkNet-53, VCG, NASNet-Large, Inception-ResNet-V2, etc Logos are converted into vectors using the trained model Vectors of logos within are stored in the database Any new logo is put through the trained model to generate a vector for similarity search against the pool of vectors Infocomm, Big Data, Data Analytics, Data Mining & Data Visualisation, Artificial Intelligence, Data Processing

Different plant species have different nutrient requirements. The current practice of urban farming uses a generic hydroponic nutrient solution that is suitable to most plant types, and a crude sensing system that only measures total ion content in the solution. This approach often results in nutrients deficiency and/or overloading and hence requires consistently monitoring. Overloading of nutrients not only increases the input costs, it also results in greater quantities of contamination in effluent to be disposed after harvest.  A targeted hydroponic nutrient solution reduces the need to periodically adjust the nutrient. The technology provider has studied and formulated different nutrient recipes that had shown improved yield compared to commercial products. This ensures the best growth for each crop type. It also reduces common problem stemming from imbalanced nutrient, e.g. leaf chlorosis due to nutrient deficiency. All these translate to a better yield and a more marketable produce for the farmers. Formulations developed include Mizuna, Kale, Lettuce, Mustard, Kalian, and Caixin. The technology provider is seeking for licensing partners from the agriculture industry. The common practice of urban farming is to discard the spent nutrient solution after a few cycles of plant growth. A targeted hydroponics nutrient solution reduces the frequency needed to periodically adjust the nutrient. The technology provider has formulated the nutrient formulation that has considered the rate of nutrient uptake by the desired plant, the nutrient’s ratio and its availability in the solution. The formulation is specific to individual crop types and ensures the best growth for each crop type, e.g. lettuce or kale.  Indoor hydroponics farmers  Fertiliser / chemical production company may wish to market this as solution to farmers or to produce as off-the-shelf products for mass market consumer Agriscience company may package this as a solution to downstream clients   Increases yield (up to 20%) Reduces the manpower needed to monitor the nutrients intake of the plant More resource-efficient with lesser nutrient adjustment Reduces common problems stemming from imbalanced nutrients Sustainable, eco-friendly solution that can potentially lead to zero-wastage Better nutrition for the consumer nutrients solution, plant growth, crop optimisation, plant nutrients, nutrients recipes, nutrients formulations, kale, caixin, mizuna, kailan, mustard, lettuce Life Sciences, Agriculture & Aquaculture, Waste Management & Recycling, Food & Agriculture Waste Management, Sustainability, Food Security

Tile grouting is the process of filling up gaps between tiles, after individual tiles have been laid onto cement screed and is a critical part of virtually every construction project. Yet, it remains a highly laborious process, and is considered one of the most physically demanding tasks as it often results in injuries to tilers' knees and back. This, in turn, leads to quality issues when grouting is not performed correctly. The construction labour shortage in Singapore, especially in the tiling/construction industry has likewise catalysed the demand for automation of such jobs - especially since such tasks are deemed to be less desirable to a younger generation of workers. This technology offer is a tile grouting robot powered by Computer Vision (CV) and Simultaneous Localisation and Mapping (SLAM) techniques, running on Robot Operating System (ROS2) to boost construction productivity and reduce the occurrence of workplace injury. The robot is able to boost productivity by at least 5 times and this results in an amortisation time of roughly 24 months. This technology offer is a compact, precise, battery-powered grouting robot that runs on Robot Operating System (ROS2) that has the following features: Grout canister with specialised nozzles to dispense and grout directly into tile gaps Grouting mechanism enables grout both centre and corner lines Self-cleaning mechanism (sponge belt) to clean up any excess grouting  SLAM techniques to automatically map out a room in real-time via two-dimensional (2D) Light Detection and Ranging (LIDAR) Recognise grout lines via a downward-facing camera with sub-millimeter precision, additionally, computer vision based detection of doors, steps, pipes and holes not visible to 2D LIDAR Grouting can be performed in any given space, including irregularly shaped rooms/corridors. Human intervention is only required to assist the robot to traverse between floors within the construction site, to re-fill the grout canister, change grout colour (if the next room has a different grout colour), and to manually fill up tile gaps that are obstructed behind pipes/objects that the robot cannot physically get to (though such cases are limited) The battery life on the robot lasts for an operating duration of 5 hours, across a floor space of 60sqm. This technology can be deployed to automate the tile grouting process for the flooring/tiling industry while the technology stack that the robot operates on can also be applied for other applications within the construction industry, such as: Tile laying Floor cleaning Quality assurance/quality checks Autonomous data collection (when outfitted with a range of sensors e.g. noise, temperature) Compared to machines/tools which are already available to aid the tile-grouting process, but still require human operation/intervention, this tile-grouting robot is autonomous and automates this laborious task. This purpose-built robot operates at a significantly lower bill of material cost when compared with non-purpose built arm effectors from large robotic manufacturers which cost more than USD$50,000 (excluding other essential sensors/components specific to a particular application) The technology owner would like to work with construction companies and tiling companies for test-bedding opportunities. Additionally, construction-related companies such as grout manufacturers, material supplies, and tool manufacturers are also of interest for potential R&D collaboration and co-development opportunities. Tile Grouting, Construction Automation, Robotics, SLAM, Robot Operating System Infocomm, Video/Image Analysis & Computer Vision, Robotics & Automation

The presence of heavy metals in food or feed powders involves contamination of the food chain and potential harm to public health, as such, rapid detection is a time-critical issue. The uncertainty about food safety caused by the possible presence of heavy metals is of concern to consumers and regulatory authorities and this is typically addressed by increasing the testing frequency of food or feed samples. However, existing testing methods are often time-consuming and require highly skilled laboratory personnel to perform the testing. This technology employs spectroscopic imaging methods and machine learning techiniques to rapidly detect heavy metals in food or feed samples. The machine learning model can perform a multi-class differentiation of the various heavy metals based on spectroscopic measurements. It is also able to predict the concentration of heavy metals present in food or feed powders using spectroscopic measurements. Minimal sample preparation is required for this method, allowing for the rapid screening of food or feed powder samples. The technology owner is interested in collaboration with companies working with food powders, with an interest in heavy metal content within food powders.    The features and specifications of this rapid screening technology include: Spectroscopic methods to collect unique spectral measurements from samples based on their chemical compositions Heavy metal classification between cadmium and lead Generation of datasets from spectral measurements to create predictive model to identify heavy metal presence and predict concentration levels Predictive model is trained on spectral measurements for increased accuracy 95% accuracy in heavy metal detection, with trace concentration detection of as low as 4ppm This technology is further customisable to include other classes of heavy metals e.g. mercury, and to include other food types e.g. seafood, meats etc. Detection and measurement of heavy metal species in food/feed powder products such as: Insect powders Animal feeds Milk powders Protein supplement powders Plant-based nutritional supplements Rapid detection of heavy metal species with minimal sample preparation Screening of large amounts of food or feed powder samples within a short period of time Model performance and prediction results are comparable to industry accepted method to measure heavy metal content  Rapid screening, heavy metals, food powders, food safety Infocomm, Video/Image Analysis & Computer Vision, Big Data, Data Analytics, Data Mining & Data Visualisation, Foods, Quality & Safety, Processes

This technology offer is a Ultrahigh Frequency (UHF) Radio-Frequency Identification (RFID) tag antenna for use on metal structures. 2 versions are available: A compact dual-band version with folded strip structure, with a total size of only 20 mm × 30 mm × 1.5 mm. This tag can be well used in different RFID systems, which work at different UHF bands, such as European and American frequencies. The reading patterns of this tag are with different directions in two bands. A single band version with a total size of only 10 mm × 30 mm × 1.5 mm. This tag can be well used in planar as well as conformal platforms, such as metallic cylinders and bearings. Automated factories should be interested in these tags, and they can use the miniaturized tags with RFID technology to intelligently detect whether the machinery and equipment are running normally. For the dual-band UHF RFID tag antenna, the main innovation is that its size is very compact. Compared to previous compact tags, this technology has the smallest dimensions of 0.06 λ × 0.09 λ × 0.0045 λ, where λ is the wavelength of free space at 915 MHz. In addition, very different from other products, the reading patterns of this tag are different in two bands. This design is able to provide a sufficiently far identification distance ( > 7 meters in European RFID band, and > 5 meters in American RFID band) at such extremely small size, proving that this tag has very high radiation efficiency. For the single band UHF RFID tag antenna, the main innovation is that its size is greatly reduce by three new strategies. Compared to previous compact tags, this technology has the smallest dimensions of 0.03 λ × 0.09 λ × 0.0045 λ, where λ is the wavelength of free space at 915 MHz. This design is able to provide a sufficiently far identification distance ( > 5 meters) with such extremely small size, proving that this tag has very high radiation efficiency. The primary application area of this technology is industrial intelligent RFID multifunctional detection system. The anti-metal tag of this system can use this technology because this single tag can cover two UHF RFID bands simultaneously. This technology also can be well used in intelligent driving RFID positioning system in different countries. Some products about logistics RFID management, luggage RFID management, containers RFID management, and shelves RFID management can be developed at different regions. The rapid development of the Internet of Things (IOTs) has spawned a variety of new technologies and product applications. The application trends of anti-metal tags are expected to be more extensive and more suitable for multi-scenario applications in RFID market. The technology can offer some business opportunities in smart RFID logistics and smart agriculture. It has very broad prospects for the rich sensing applications of the IOTs in the near future. This design can save a lot of installation space for the rest of the industrial equipment, especially for the dual band tag which works at 2 different frequencies, with its multifunctional radiation beams. Due to its highly efficient radiation, it can reduce the input power of the transmit components of its RFID system. Other risk factors due to the continuous heating state of the system can be further reduced.  The technology owner is keen to license this technology to RFID application technology companies, including for logistics management, access control and equipment management, etc. RFID, UHF, RFID for Metal Electronics, Sensors & Instrumentation, Embedded Systems

This technology offer is an integrated, CMOS-compatible, compact device that provides dispersion compensation of dispersion in optical fibers. Dispersion impairments is a well-known problem in the transmission of high-speed data over fiber, that limits both the fiber reaches, or poses lower limits on the power required. The technology developed allows a seamless, very low loss method for compensation of fiber dispersion, providing high magnitudes of dispersion for countering dispersion in optical fibers. Without dispersion compensation, signals are susceptible to degradation from optical fiber dispersion, with the extent of degradation worsening with longer fiber reaches. Without proper dispersion compensation, transmitted data will experience high Bit Error Rates (BER) at the receiver. This technology solves this important problem and increases the fiber reaches which may be served.   The technology has the following specifications: Dispersion magnitude is scalable through appropriate design, depending on the fiber lengths that need to be compensated for. Operating wavelength is tailorable. Tunability may be introduced through thermo-optic control. Dispersion compensation is applicable for both intensity modulated direct detection modulation formats and coherent modulation formats. This device is CMOS-compatible, low loss and operates in transmission mode. May be seamlessly integrated with photonic integrated circuits. The technology owner has experimentally demonstrated that the technology works. High-speed characterization using 30 Gb/s NRZ and 30 Gbaud/s PAM4 data showed a restoration of the eye diagram that deteriorated after propagating through 2km of optical fiber. BER characterization showed a 1.3dB improvement in power penalty out of a 1.8dB degradation at the error-free (BER = 10-12) level. Scalable dispersion has also been experimentally proven. The transmission of high speed data over optical fiber is well known to be impaired by dispersion in the optical fiber. This technology provides a very low loss solution to dispersion compensation and has been shown to restore the eye diagram and improve the bit error rate of high speed data. Potential applications would be pre- or post- dispersion compensation of optical fiber dispersion. Transceivers which serve long fiber reaches and/or utilize high speed data could incorporate this technology in the transceiver chip (either transmitter or receiver), to provide an integrated, low loss, CMOS-compatible means of high quality dispersion compensation that can be easily integrated with the rest of the transceiver chip. The silicon photonics transceiver market is projected to grow to S$6.4 billion by 2026 with a compound annual growth rate of 25%. Asia Pacific is expected to show the fastest growth with increasing adoption of high-speed data systems, supportive government initiatives, and fast-growing consumer demand. Increases transceiver reaches Provides either pre- or post- dispersion compensation, with very low loss. This would allow existing transceiver products to serve longer reaches, or support higher data rates. It would also open up new opportunities for new product lines which provide integrated dispersion compensation, reducing the complexity of deployment for network operators. Very low cost, requiring only integration of the device within the transceiver chip, negligible increase in power budget. The technology owner is looking for R&D collaboration, IP licensing/acquisition, testbedding opportunities with optical transceiver companies, data center hardware companies, telecommunications companies or silicon photonics companies. Electronics, Semiconductors, Memory & Storage