4th ACE Call Awards

A portable scent screening system for food freshness monitoring

PI: Chen Xiaodong (MSE)
Ling Xing Yi (SPMS); Zheng Yuanjin (EEE); Balázs Zoltán Gulyás (LKCM)

Artificial scent screening systems (known as electronic noses, E-nose) have been researched extensively. In food industry, scent screening is an effective technology for food quality monitoring and food safety surveillance. Few Enoses have been integrated with food supply chains because they suffer from either sensing (high temperatures or cumbersome power supply and wiring) or pattern recognition issues (low accuracy in analysis of non-linear, multidimensional datasets). To commercialize E-noses, we need a system that has both a robust cross-reactive sensor array and a data analysis platform that can extract information from non-linear datasets and accurately predict fingerprint patterns.

To solve the systematic problem of E-nose in food industry, we need technique breakthrough in both sensor development and data processing fields. To build an integrated system, interdisciplinary cooperation is highly needed, especially researchers from material science, device fabrication, as well as electrical engineering and computer science.

In the sensing technology, we intend to use cross-reactive colorimetric barcode combinatorics. The designed sensing bars are highly cross-reactive and perform well at room temperature. Barcode combinatorics produce 3n (where 3 represents RGB and n is number of bar) non-linear, multidimensional datasets, which provides a unique scent fingerprint. In the data processing technology, we plan to use deep convolutional neural networks (DCNNs), an end-to-end network with several nonlinear activation functions, which is suitable for non-linear multidimension data analysis. Based on a fully supervised DCNN trained using labelled barcode images, it is expected to obtain a high prediction accuracy. Moreover, incorporating DCNN into a smartphone application will help to build a simple platform for rapid barcode scanning and food freshness identification in real time. By combining cross-reactive colorimetric barcode and DCNN technology, we aim to build an artificial scent screening system that concurrently provides scent fingerprint and fingerprint recognition. Based on barcode images for meat freshness sensing, our vision is to build a portable meat freshness monitoring system.

Towards a green future: improving urban farming under tropical weather conditions, for locally-grown vegetables of high quality

PI: Ma Wei (SBS)
Co-PI: Huang Changjin (MAE); Jimmy K. Hsia (MAE); Cheng Wei Ning, William (SCBE); Shirley S. Ho (WKWSCI)

Green vegetables, which are important food resource and essential components in urban farming, play pivotal roles in human nutrition and health. Recent advances in understanding plant lipid metabolism have brought new insights into how the membrane lipid remodelling mediates the function of chloroplast, the factory organelle in plant cells, in response to diverse environmental stimuli. However, little is known about the composition changes of polar lipids in leafy vegetable membrane as environmental conditions vary, and how they alter the function of chloroplasts at cellular level and consequently the mechanical properties of vegetable leaves at tissue level remains largely elusive. We aim to investigate their relationships under various environmental conditions. We are particularly interested in conditions of increased humidity, temperature and light intensity, as these factors are typical features of the tropical climate in Singapore. We hypothesize that changes of environmental conditions regulate the nutritional and texture qualities of vegetable leaves by mediating chloroplast structure and function. The knowledge gained from this study will provide us a better strategy to improve vegetable growth and nutrition for urban farming.

The proposed research is appropriate for ACE program as our approaches go beyond the conventional methods, such as identifying new genes/pathways or generation of genetically modified (GM) vegetables, which have been broadly used in regular biological researches that are appreciated and supported by traditional competitive funding schemes. This proposal represents an interdisciplinary effort to help optimize urban farming conditions from a new perspective by integrating biomechanics and computational simulations with plant biochemistry and physiology. We are confident that the data collected and analyses conducted here will allow us to identify the optimized environmental conditions that will make leafy vegetables irresistible to consumers because of their enhanced nutritional and texture qualities. The outcome from our proposed research is not only scientifically important, but also has broad economic and societal impact.

Unlocking human superpowers through AI-aided health and wellness model to enhance sleep quality

PI: Josip Car (LKCM)
Co-PI: May O. Lwin (WKWSCI); Theng Ying Leng (WKWSCI/ARISE); Chia Yong Hwa Michael (NIE); Vivien Huan (NIE); Ho Moon-ho Ringo (SSS/ARISE); Joty Shafiq Rayhan (SCSE)

Majority of chronic diseases could be prevented by changes in health behaviours – amongst them by improving sleep. “Sleep is a superpower”. The ability to restore cognitive, emotional and physical energy through sleep is essential for maintaining good health, performance and well-being. Yet it is underutilized by majority. Singaporeans are the second most sleepdeprived nation in the world, with 62% not getting enough sleep.2 The fact that small improvements in sleep at the population level can be amplified into much larger economic gains is something of a wake-up call.

Insufficient length and poor quality of sleep have multifarious reasons and route-couses, not least lifestyle choice. At student age these are not deep-seated insomnia or a serious disease which requires medical treatment but rather require an evidence-based practical advice, dispelling of myths, and coaching for positive behaviour activation; ameliorable to major improvement apart from those where mental health problems may be uncovered (which is equally highly beneficial).

The aim of this interdisciplinary research, involving medical, population health, behavioural, communication, and computer scientists and experts, is to design, develop, and evaluate a scalable, sustainable, high-fidelity at scale and highly cost-effecitve, innovative solution which can take sleep as an exemplar for a wide range of health and wellbeing interventions. This grant will unlock the path to creation of a comprehensive human-AI symbiosis health coach. We will leverage Healthy Campus app, conversational agents, behaviour change techniques from nudging to behaviour activation and automated response including based on wearable sensor data. Good sleep student peer health coaching will achieve NTU students’ better academic performance and health outcomes. As the study is driven by design thinking, is interdisciplinary and serves as a proof of concept of a bigger health coaching paradigm, it will specifically benefit from ACE funding.

Robots as Medical Professionals: Unpacking and Mitigating Public’s Resistance towards Rehabilitation Robots

PI: Sharon Ng (NBS)
Co-PI: Domenico Campolo (MAE) 
Collaborator: Gabriel Aguirre Ollinger (ARTICARES Pte Ltd); Wee Seng Kwee (Tan Tock Seng Hospital)

The use of rehabilitation robot is becoming increasingly popular as an alternative form of treatment in recent years. The recent COVID epidemic showcased the value of using robots in a medical setting. However, scant research has examined the human aspects of such technology. A handful of research has shown that people are generally resistant to the use of robots in the medical setting (Longoni, Bonezzi, and Morewedge 2019). Why is this so? How do people perceive and interact with robots in a medical setting? The research presented in this proposal aims to fill this gap in the literature.

Focusing on the use of robots in upper-limb rehabilitation, we aim to examine if the use of robot (versus human) therapist would 1) affect the extent they would persevere through a treatment, 2) influence their anticipated and reported sensation of pain and 3) shift patients’ satisfaction with the treatment. This is because robot therapists and human therapists evoke different levels of expectations of treatment effectiveness, empathy and mindset. This research further proposes scalable interventions that may help to mitigate any negative perception about therapeutic robots. If we can enhance patients’ acceptance of therapeutic robots by addressing their concerns about such technology, we can change the whole concept of movement rehabilitation. The aim is to give patients the freedom to get high quality therapy exercise in the comfort of their own home and in their own time, all while receiving expert monitoring and feedback from a remotely based clinician.

the cross-disciplinary nature of this proposal, the niche context we are focusing on (i.e., rehabilitation, pain sensation), and the exploratory nature of our predictions, we feel that this project would be a better fit with ACE programme that values bold and interdisciplinary ideas. The multidisciplinary nature of this proposal does not fit the traditional funding schemes. The proposed ideas fit the theme on “Responsible Technology” highlighted in this grant call. This proposal also is in line with NTU’s objective to bring robots a step closer to the public and our findings would help to understand the bottleneck behind robot acceptance by the general public.