Innovation Challenges

How might we implement an integrated end-to-end Guest Recognition AI solution at the critical guest touch points to improve guest experiences?

In the hospitality industry, creating personalised and seamless guest experiences is key to fostering loyalty and increasing customer satisfaction. As a PARKROYAL COLLECTION hotel we seek to improve operational efficiency and provide guests with a personalised experience and we believe technology such as facial recognition offers a compelling opportunity. Facial recognition can streamline check-in processes, enhance personalised greetings, and help staff anticipate guest needs by identifying them upon arrival.

This challenge seeks a solution that leverages facial recognition technology to optimise guest interactions, while ensuring compliance with data privacy laws and easy integration with our existing hotel management systems such as Opera and (mobile) check-in platforms. The right solution would not only improve guest satisfaction but also reduce operational burden on staff, offering a more streamlined, cost-effective, and personalised hotel experience.

We are looking for a solution that can be deployed at the following critical touchpoints:

• Hotel Reception: Receptionists able to greet guests by their last name upon arrival.
• Restaurant: Food & Beverage personnel able to greet guests and identify breakfast entitlement, be aware of allergies and dietary restriction of our guests.
• Club lounge, Spa and Wellness: Personnel able to greet guests, guest access and eligibility to be known upon arrival based on membership level of our loyalty programs or booking information.

We are looking for a solution that respects the data privacy of our guests and doesn’t store more information than strictly necessary, integrates with legacy systems and is cost-effective. Identification, of course, needs to be real-time and be able to extract data from multiple systems such as the reception- and mobile check-in systems, loyalty program database, passport scanner, and others.

Currently, no off-the-shelf solution fully meets our needs. Existing systems either lack the accuracy, necessary integration capabilities or fail to comply with stringent data privacy regulations.

What We Are Looking For

Technical Requirements:

• Facial Recognition Accuracy: The system must be able to accurately identify guests upon arrival with minimal errors, even in high-traffic areas.
• Integration with Existing Systems: The solution must integrate seamlessly with systems like Opera (hotel management) and mobile check-in platforms, enabling automatic updates of guest data.
• Data Privacy and Security: The solution must ensure compliance with local privacy regulations, such as PDPA, by collecting minimal guest data (e.g., last name, photo, eligibility, allergies) and securely storing or processing this information, preferably using on-premise systems. If cloud based solutions are used, they must be secure.
• Operational Efficiency: The system should reduce the need for manual data entry and help staff serve guests faster by automatically pulling up relevant guest information.
  

Performance Requirements:

• Guest experience: The main driver for this challenge is to improve the guest experience. Solutions will therefore be evaluated based on their impact on the experience as evidenced by being mentioned in guest reviews (for example, mentioning the greetings by last name at check in and dietary awareness) as well as an uptick in review scores from our guests.
• Cost effectiveness: Please see below section

Cost Target

Cost targets will be determined on a case-by-case basis. In general we evaluate solutions based on their return on investment but the specific target will depend on the solution. Separate from the investment needed for the system we have a cost target of S$10,000 per year, per hotel for operational expenses.

Timeframe for Development

Phase 1: POC development: Q2- Q3 2025 at our PARKROYAL COLLECTION Pickering hotel in Singapore.

Phase 2: Commercial roll-out: to be determined on a case-by-case basis.

Potential Market / Business Opportunity

If the solution is successful, PARKROYAL COLLECTION Pickering, Singapore is willing to support a further roll-out across other locations in Singapore. Furthermore PARKROYAL COLLECTION Pickering, Singapore is part of the Pan Pacific Hotels Group with about 50 hotels in our portfolio where the solution can also be implemented.

Additionally, we are open for solution providers to deploy their solution with other players.

Resources

• Awarded innovator will receive support and collaboration opportunities for pilots with a potential financial support of up to S$50,000.

• Mentorship and support for solution development.
• Access to relevant datasets, and pilot opportunity at our PARKROYAL COLLECTION Pickering, Singapore.

• Up to S$20,000 grant support from EnterpriseSG.

Other Considerations

We are looking for SMEs and startups with solutions that can be implemented in a relatively short time frame (TRL of 5 and higher).

For Background IP (BIP), both parties will retain their respective IPs brought into the project. In the event of new Foreground IP (FIP) creation, ParkRoyal is agreeable to the FIP being retained by the solution provider.

Learn More

How might we develop an AI-based detection and analytics systems to identify key yard performance parameters to help us improve safety, productivity and efficiency?

As an EPC company, for Seatrium tracking what is going on in our construction sides is important since it helps us to monitor key safety, productivity and efficiency parameters. Currently, Seatrium works with many different large contractors in different yards across the world. These contractors can include many different parties such as welders, cable layers etc. Currently, we have an existing practice where our safety managers and or project managers will conduct weekly (or sometimes daily) walkarounds to ensure work is done in compliance with the relevant safety, productivity and efficiency requirements and standards. While doing these manual checks can be sometimes useful, there are several limitations to their effectiveness including:

• Consistency and Accuracy: Supervisors can be inconsistent in their detection abilities, leading to variability in results. Factors like fatigue, attention, and experience can affect accuracy.
• Scalability: As the volume of workforce to be detected is very large, manual methods become impractical. Automated systems can handle large datasets more efficiently.
• Subjectivity: Manual detection can be subjective, with different people potentially interpreting or detecting things in different ways.
• Labour Costs: Relying on Safety/QA/Ops manpower for detection tasks can be expensive and resource-intensive compared to automated solutions

• Track if workers wear their PPE vests and helmets
• People counting
• Checking the productivity and general presence of workers
• Checking the occurrence of sparks or flames when welding
• etc

We are looking for solutions that can be integrated across our yards and are able to provide datasets and data integrations across different devices including (CCTV) camera’s, robots dogs, smart glasses etc.

What We Are Looking For

Technical Requirements:

We would like to be able to track the following use cases:

1. Yard activities – grinding, welding, cutting, cable pulling, scaffolding etc
2. Yard Safety risk – drop objects, tripping, sparks, flame, fire, wet surface
3. Productivity evidence – Weldment defects and completion, tag recognition, pressure recorder and gauge, sensor reading
4. Of the aforementioned use cases, we would expect a solution provider to be able to implement at least 3-4 use case
5. Ideally, the solution would also provide an integration with LLM’s so that our staff can interact with the system by using natural language
6. We would expect the system to be able to provide real-time intervention for certain (safety related) use cases
7. Ideally, the solution should include a dashboard to monitor key KPI’s for safety, quality and productivity. If this is not feasible, we are open for this to be included in a further commercial rollout post POC
8. The solution should eventually be able to be deployed on a variety of IOT devices such as smart glasses, robot docs, smart cameras etc
9. The data source of these are from the various IOT devices such as smart glass, robot dogs or smart cameras. The AI detection must be able to provide analytic insights for process improvements
10. We are looking for a solution that can provide us operational insights to improve our operations

Performance Requirements: 

The business performance of solutions will be evaluated on a case-by-case basis. However, generally speaking, we would be looking for the following business benefits:

• We are looking for an efficiency gain of 20% more in terms of productivity increase
• We would be looking to save approximately S$100-200K on personnel doing ad-hoc patrolling

Cost Target

Cost targets will be determined on a case-by-case basis.

Timeframe for Development

Phase 1: POC development: Q3-Q4 2024.

Phase 2: Commercial roll-out: to be determined on a case-by-case basis, target implementation by Q1 2025

Potential Market / Business Opportunity

If the solution is successful, Seatrium is willing to support further deployment across other yards worldwide. Seatrium has businesses all over the globe. Seatrium main customers include major energy companies, vessel owners and operators, shipping companies, and cruise and ferry operators. Seatrium operates shipyards, engineering & technology centres and facilities in Singapore, Brazil, China, India, Indonesia, Japan, Malaysia, the Philippines, Norway, the United Arab Emirates, the United Kingdom and the United States.

Resources

Cash contributions:

Up to S$50,000 to support the POC development. Note that the POC development budget will be dependent on the quality of the solutions provided and will be committed only based on the quality of the specific POC proposal.

In-kind contributions:

• Mentorship and support for solution development.
• Access to relevant datasets, lab facilities and pilot site(s).
• Access to relevant visual data (e.g. CCTV recordings, access to the actual construction site and existing remote IOT infrastructure

Additional contributions from EnterpriseSG:

Up to S$20,000 grant support from EnterpriseSG.

Other Considerations

We are looking for SMEs and startups with solutions that can be implemented in a relatively short-time frame (TRL of 5 and higher). For Background IP (BIP), both parties will retain their respective IPs bought into the project. In the event of a new Foreground IP (FIP) creation, FIP ownership will be discussed on a case-by-case basis.

Learn More

How might we use AI and computer vision technologies to understand human patterns and use these to improve safety and productivity in warehouse operations?

Toyota Tsusho is a large multinational trading and supply chain management company part of the Toyota group. This challenge is part of the global parts and logistics division responsible for the supply chain of automotive parts.

Warehouse environments are often prone to safety risks, such as near-misses, damages to goods, and accidents involving workers, goods, or infrastructure. In addition to safety concerns, the productivity of operations, including packaging line efficiency and space optimization, can significantly impact operational costs and lead times.

The focus of this challenge is to leverage innovative AI and computer vision technologies, to gather and analyze real-time data to enhance safety, streamline processes, or improve the overall efficiency of warehouse operations. The technology should further serve as platform for further automation in logistics, increases in operational productivity, and improved productivity.

What We Are Looking For

Technical Requirements:

• We are looking for a decrease in damages and incidents of 20% or more.
• We are looking for a productivity gain of 10% or more for the packaging line (speed / reduction in wrong packaging resulting in reduction of redelivery)

Cost Target

Cost targets will be determined on a case-by-case basis.

In a post proof-of-concept phase and a solution that is implemented over several of our warehouses, we are looking for a cost target of not more than S$50 per video stream per site for a total of S$2500 per month cost (running cost).

Timeframe for Development

Phase 1: POC development: Q2- Q3 2025.

Phase 2:  Commercial roll-out: to be determined on a case-by-case basis

Potential Market/ Business Opportunity

If the solution is successful, Toyota Tsusho is willing to support a further roll-out across our 100-200 sites in the Asia-Oceania region. Additionally, we are open for solution providers to deploy its solution with other players.

Resources

Learn More

How might we use the power of AI to unlock our data and enhance our guest experience with proactive, personalised interactions?

In a competitive hospitality market, providing a personalised and seamless guest experience is critical for enhancing guest satisfaction, loyalty, and operational efficiency. At ParkRoyal Hotel, guest data is collected across various touchpoints, from check-in and F&B interactions, room service, spa appointments, and guest surveys. However, this data is currently fragmented across 125 systems. This creates challenges in providing guests with an optimal experience as our ability to recognise VIP guests, anticipate special requests, or act on past feedback is limited. Therefore, we are looking for a solution that combines (unstructured) data from many sources to enable a proactive way of handling the guest experience.

An example of a guest experience we would like to improve is the check-in. Besides being able to greet our guests by name (refer to our other challenge statement), we want to be able to, for example, refer to the comments they made after their previous visit. These comments from the guest survey or feedback are matched to their guest profile, and we are alerted before arrival so we can prepare the room according to their preference. A pop-up shows up during check-in so we can highlight how we acted upon their feedback during check-in and improve the guest’s satisfaction. Some other touchpoints we would like to improve are in the driveway, lobby, F&B outlets and spa.

A second goal of the challenge is to enable proactive, suggestive selling methods. The solution should provide relevant products and services for our guests. This should be based on their guest profile and patterns detected in our data by the solution. This should enable us to increase revenue while delighting our guests during the entirety of their stay.

Some examples of the systems we currently use are Opera, PMS, POS, Stayplease, Vingcard Zigbee, Positioning tracking through BLE, UHF RFID, GRMS, FCU and Controls. The goal of this challenge is not to integrate all this data but to use novel machine learning (ML) and generative artificial intelligence (genAI) to improve the guest experience based on data patterns discovered. This should improve our institutional memory to transform our business from reactive to proactive.

What We Are Looking For

Technical Requirements:

• Real-time insights: the solution should leverage the data available across systems to provide real-time, proactive insights for the various departments and guest touchpoints
• Operational integration: the output of the solution should be in the form of a pop-up on desktop applications (reception, spa), a pop-up on iPad (breakfast, food and beverage), and by voice in staff earpieces (driveway, lobby).
• Privacy by design: The solution must ensure compliance with local privacy regulations, such as PDPA, by collecting minimal guest data (e.g., last name only, not full name)
• Data access: ability to obtain data from many different systems in various ways, either through API, middleware, database access, text files, etc.
• Security: many of the data in our systems are cloud-based, so we are open to cloud-based solutions as long as they are securely designed
• Operational Efficiency: The system should reduce the need for manual data entry and help staff serve guests faster by automatically pulling up relevant guest information.

Performance Requirements:

• Guest experience: the main driver for this challenge is to improve the guest experience. Solutions will, therefore, be evaluated based on their impact on the experience as evidenced by being mentioned in guest reviews (f.e. mentioning the pro-active services) as well as an uptick in review scores from our guests.
• Revenue increase: suggestive selling should increase the revenues per stay
• Cost-effectiveness: please see below.

Cost Target

Cost targets will be determined on a case-by-case basis. In general, we evaluate solutions based on their return on investment, but the specific target will depend on the solution. Separate from the investment needed for the system, we have a cost target of S$10,000 to $12,000 per year, per hotel for operational expenses.

Timeframe for Development

Phase 1: POC development: Q2- Q3 2025 at our PARKROYAL COLLECTION Pickering hotel in Singapore

Phase 2: Commercial roll-out: to be determined on a case-by-case basis

Potential Market/ Business Opportunity

If the solution is successful, PARKROYAL COLLECTION Pickering, Singapore is willing to support a further roll-out across other locations in Singapore. Furthermore, PARKROYAL COLLECTION Pickering, Singapore is part of the Pan Pacific Hotels Group with about 50 hotels in our portfolio where the solution can also be implemented. Additionally, we are open for solution providers to deploy their solution with other players.

Resources

• Awarded innovators will receive support and collaboration opportunities for pilots with potential financial support of up to S$50,000.

• Mentorship and support for solution development and or access to relevant datasets, and pilot opportunities at our PARKROYAL COLLECTION Pickering, Singapore.

• Up to S$20,000 grant support from EnterpriseSG on a matching basis.

Other Considerations

We are looking for SMEs and startups with solutions that can be implemented in a relatively short time frame (TRL of 5 and higher).

For Background IP (BIP), both parties will retain their respective IPs bought into the project. In the event of new Foreground IP (FIP) creation, ParkRoyal is agreeable to the FIP being retained by the solution provider.

Learn More

How might we preserve and enhance biodiversity through AI-driven solutions in urban planning, development, and maintenance?

As the world’s trusted infrastructure consulting firm, we recognize that biological diversity and healthy natural ecosystems are fundamental to human well-being and economic prosperity for all people — and will prove crucial in addressing climate change and environmental challenges.

Biodiversity exhibits a high degree of natural variability, with populations of plants and animals changing through space and time in response to a myriad of factors such as climate, resource availability, and populations of predators and prey. Traditional ecological approaches rely on field surveys to characterise biodiversity resources. To account for this natural variability, these surveys require significant time and cost inputs, particularly for seasonally occurring species (i.e., migratory avifauna) or rare, cryptic species (e.g., Eurasian Otter).

Urban developments can result in significant biodiversity loss due to the conversion of natural habitats into built environments, resulting in habitat loss, fragmented green spaces, reduced habitat quality and species diversity. Due to the natural variability of biodiversity data, relying on traditional ecological surveys means that:

• The impacts of urban developments on biodiversity resources are difficult to predict;
• The effectiveness of nature-based solutions in urban settings cannot be accurately assessed.

AECOM is therefore looking for solution providers that can provide solutions to leverage AI to monitor, analyse, and optimise biodiversity in the planning and management of urban environments, enabling ecologists, urban planners and developers to create more sustainable, nature-friendly environments.

We believe that this solution can help us to:

• Develop a novel product offering where we provide faster and better biodiversity insights to our customers
• Safe costs by doing away with costly and inaccurate surveying

Previous approaches have included isolated surveys and periodic assessments using traditional ecological methods. These often provide a snapshot of conditions rather than continuous monitoring, making it difficult to track changes in biodiversity over time and differentiate between natural variations and those resulting from urban development. Manual data collection and analysis are resource-intensive and lack the ability to predict the impact of urban development on ecosystems effectively. Additionally, many solutions fail to integrate with existing urban data platforms or do not scale well to large urban areas. We have not tried AI solutions previously.

We are not keen on solutions that focus solely on traditional ecological surveys without leveraging AI for data analysis or automation. Additionally, solutions that only provide generic recommendations without being tailored to specific urban contexts or fail to integrate with urban planning processes would not be suitable.

Finally, it is important to note that AECOM is willing to support solution providers in terms of data availability. We are agnostic in terms of the tools that are being used and are open to the use of IOT sensors, computer vision-related solutions and simulations.

What We Are Looking For

Technical Requirements:

• The solution should be compatible with existing systems such as GIS and iBAT
• Most likely we would first like the system to focus on Hong Kong and Singapore as a first pilot
• The solution should be able to process and analyse data from diverse sources such as satellite imagery, digital sensors, and citizen science inputs.
• The solution should be able to perform real-time monitoring of biodiversity metrics (e.g., species occurrence, abundance and richness; habitat value and connectivity).
• The solution should be able to perform predictive capabilities to assess the impact of urban development scenarios and nature-based solutions on biodiversity
• We are looking for a user-friendly interface for ecologists, urban planners and decision-makers to visualise data and receive actionable insights
• We would like to have reporting interfaces that can be plugged directly into international reporting standards like TNFD
• An integration with LLMs would be a nice to have so that users can interact with the system using natural language
• The solution should adhere to data privacy standards if collecting data from citizen science efforts or through mobile applications 

Performance Criteria:

Solutions will be evaluated on a case-by-case basis. Generally, we are looking at the following performance criteria:

• Reduction of survey efforts and labour cost
• A better and more unique offering towards our customers
• Overall system ROI

Cost Target

Cost targets will be determined on a case-by-case basis.

AECOM is willing to support a POC through a combination of in-kind and cash investment but does require the startup to co-invest as this is a co-innovation project.

Timeframe for Development

Phase 1: POC development: Q3-Q4 2025.
Phase 2: Commercial roll-out: To be determined on a case-by-case basis, target implementation by Q1 2026.

Potential Market/ Business Opportunity for the Product/ Solution

If the solution is successful, AECOM is willing to support further deployment across other locations worldwide. We believe the solution could be extended to other urban areas, those with active biodiversity and sustainability initiatives, smart city programs, and climate adaptation strategies would be primary targets. AECOM might be requesting for a period of exclusivity to bring the solution to AECOM clients first.

Resources

• Up to S$50,000 budget to support pilot development. POC development budget will be evaluated on a case-by-case basis depending on the quality and feasibility of the proposed solutions.

• Mentorship and support for solution development.
• Access to relevant datasets, lab facilities and pilot site(s)

• Up to S$20,000 grant support from EnterpriseSG on a matching basis.

Other Considerations

We are looking for SMEs and startups with solutions that can be implemented in a relatively short-time frame (TRL of 5 and higher).
Intellectual Property (IP): For Background IP (BIP), both parties will retain their respective IPs bought into the project. In the event of new Foreground IP (FIP) creation, FIP ownership will be discussed on a case-by-case basis

Learn More

How do we leverage AI-driven solutions to create, measure, and optimise the social value (such as community benefits, healthcare, living conditions and social equity) generated by infrastructure project?

As the world’s leading infrastructure consulting firm, our clients come to us for solutions that meet their ESG targets. While the environmental and governance parts are well developed the social part is not as developed yet. Governments planning, developing and upgrading large infrastructure projects such as cities, neighbourhoods, transportation systems, utilities, and public spaces are looking for solutions to effectively measure the social value during the planning, execution, and operational phases.

Large infrastructure projects generate social value, such as community well-being, social cohesion, accessibility, and economic opportunities. Current evaluation methods often rely on subjective assessments, such as surveys, Therefore, it is currently challenging to make informed design decisions during the planning phase and have an objective measure of the social value impact of that decision. As an example, decisions that we are looking to optimise are: where to place a hospital or school, what should the square metre size be, and what type of healthcare is needed given the surrounding social conditions and demographics.

To make these types of decisions, we are looking for a tool that uses publicly available data and location-specific data sets that calculates the social value impact of design decisions in real time during the planning phase. The solution should ideally express the social value as an objective number so decisions can be compared. Furthermore, the tool should also be able to measure the actual social value after project completion using all data available (including a.o. survey data and social media).

The methods currently used most are pre- and post-construction surveys, community consultations, and economic impact studies. The nature of the questions in the surveys is qualitative and subjective. Making the data more objective is part of the goals of this challenge. Some indicators we currently use for expressing social value are savings in public health costs, job creation, community well-being and educational outcomes. We are open to other data and indicators for use in creating the social value score. Besides expressing the social value as a score, the ideal solution should also make design recommendations to improve the social value created.

We are not interested in solutions that focus exclusively on economic impact assessments without considering broader social impacts, or those that are not scalable or cannot provide insights throughout the project's lifecycle. Additionally, any solution that lacks AI-based data analysis and relies solely on manual data collection will not meet our needs.

What We Are Looking For

Technical Requirements:

• Pattern recognition: The AI models should be capable of identifying and analysing patterns in community sentiment and well-being related to infrastructure projects.
• Predictive: the solution should use predictive analytics to estimate potential social value outcomes during the planning stages.
• Subjectivity: Social value is a subjective measure and different databases use different values for economic- and social opportunities for example. The solution should pick the best from what’s available and customise for each situation but justify why particular indicators have been selected
• Real-time: The solution should provide (near) real-time feedback on the impact of design decisions on the social value created (in the planning phase). This includes social value dashboards for stakeholders, enabling visualisation of metrics like community satisfaction, accessibility improvements, and economic inclusion.
• Automatic data collection: data collection should be automatic through integration with digital platforms (e.g., government’s open data initiatives, social media, mobile applications, etc.).
• Data privacy: Compliance with data privacy regulations, including anonymisation of community feedback and adherence to local and international privacy laws.

• Performance criteria for the solutions will be evaluated on a case-by-case basis.
• Ideally we should be able to take the tool to our clients as part of our go-to-market.
• We are open to exploring different business models together with solutions providers to not only use the solution in our own work but to also make the solution available to our clients as part of our broader offering. Depending on the business model chosen, the performance criteria will vary.

Cost Target

Cost targets will be determined on a case-by-case basis.

Timeframe for Development

Phase 1: POC development: Q3-Q4 2025.

Phase 2: Commercial roll-out: To be determined on a case-by-case basis, target implementation by Q1 2026.

Potential Market/ Business Opportunity

If the solution is successful, AECOM is willing to support further deployment across other locations worldwide. We believe the solution could be applied to a wide range of infrastructure projects in Asia and globally to help infrastructure providers prioritise projects according to needs. The solution also has potential in adjacent markets like real estate, urban development agencies, and community-focused organisations to ensure social equity, get project buy-in and avoid costly delays.

As stated under performance criteria, we are open to exploring different business models together with potential solution providers and leverage our global network to create a win-win for everyone involved.

AECOM might be requesting for a period of exclusivity to bring the solution to AECOM clients first.

Resources

• Up to S$50,000 budget to support pilot development. POC development budget will be evaluated on a case-by-case basis depending on the quality and feasibility of the proposed solutions.

• Mentorship and support for solution development.
• Access to relevant datasets, lab facilities and pilot site(s)

• Up to S$20,000 grant support from EnterpriseSG on a matching basis.

Other Considerations

We are looking for SMEs and startups with solutions that can be implemented in a relatively short-time frame (TRL of 5 and higher).

Intellectual Property (IP): For Background IP (BIP), both parties will retain their respective IPs bought into the project. In the event of new Foreground IP (FIP) creation, FIP ownership will be discussed on a case-by-case basis

Learn More

How might we leverage AI-driven solutions to enhance the climate resilience of urban infrastructure, enabling better adaptation to extreme weather events and long-term climate impacts?

Urban infrastructure is increasingly exposed to climate risks such as rising sea levels, extreme heat, intense rainfall, and storms. Traditional approaches to climate resilience often rely on static models and historical data, making them inadequate for predicting and adapting to the evolving nature of climate change. Coastal defences, drainage systems, and building codes are all based on these models and are often not adapted to these risks. Even the most advanced models do not optimally support decision making in what to build in defence of these risks, that’s the main goal of this challenge. We are, therefore, looking for a (AI-driven) solution to assess vulnerabilities, predict local climate risks, and optimise the design, maintenance, and retrofitting of infrastructure to improve climate resilience.

Current practices include computer models and expert meetings to discuss risks, probabilities, and opinions on potential solutions. The accuracy of these methods for predictions on the level of the specific infrastructure element is limited. Plus, current processes are often reactive to the actual climate risks.

Predictive models are currently unable to process real-time weather data (so the models are always up-to-date) and link the probability of climate risks to the performance of specific infrastructure elements. Digital twins developed for climate resilience have so far failed due to limitations in data integration and the lack of scalability to the level of larger urban areas. Limitations in data integration do not relate to the data not being available but rather to integrating the data in a way that creates a comprehensive overview, a complete story of the risk so it delivers actionable insights in how to prioritise potential infrastructure adjustments.

This challenge seeks a solution that both increases the accuracy of the climate risk prediction models for infrastructure, but mainly a solution that suggests and assesses the best way to protect communities and assets against these risks. When a good risk prediction is currently available it is still unclear what infrastructure to build in response to the specific risk, such as coastal walls, drainage systems, reservoirs and elevated roads. The solution should make it easy for infrastructure planners and asset owners to quickly assess alternatives. Furthermore, with a solution available the assessment can be performed more frequently and at lower costs, based on the updated data.

The model should incorporate the vast amounts of data available, such as governmental open data, weather data, historic datasets, satellite, asset types, asset values, existing sensor data, and hydraulic models.

The ideal solution should help us and our governmental clients understand the actual climate risks for their infrastructure, support in the infrastructure planning phase and with monitoring the risks on a frequent basis. The solution should also support our large real-estate asset owners in understanding the impact of climate risks on their portfolio.

Solutions that focus only on monitoring weather patterns without linking them to specific infrastructure impacts, or those that cannot adapt their models over time to reflect changing conditions, would not be of interest. Similarly, solutions that rely solely on manual data input and do not utilise machine learning or AI to process large datasets will not meet the challenge's requirements.

What We Are Looking For

Technical Requirements:

• Compatibility: with existing infrastructure data platforms such as [include relevant examples] and climate data sources (e.g., satellite data, weather stations).
• Infrastructure: Cloud-based with options for edge processing to enable real-time risk assessment for remote or critical infrastructure.
• Scalability: Scalable to different types of urban infrastructure, such as transportation networks, water supply and drainage systems, and energy grids.
• Real-time: The solution should be capable of evaluating adjustments to infrastructure plans in real-time, based on up-to-date climate data and infrastructure performance.
• Predictive capabilities: To forecast extreme weather impacts on infrastructure over different time horizons (e.g., immediate, short-term, long-term).
• Suggestions: The solution should support in decision making by recommending adaptive measures (e.g., prioritising maintenance, retrofitting suggestions, building new somewhere else).
• Integration: Integration with city management dashboards to provide actionable insights for city planners and infrastructure managers.
  

Performance Criteria:

• Performance criteria for the solutions will be evaluated on a case-by-case basis.
• Ideally we should be able to take the tool to our clients as part of our go-to-market.
• We are open to exploring different business models together with solutions providers to not only use the solution in our own work but to also make the solution available to our clients as part of our broader offering. Depending on the business model chosen, the performance criteria will vary.

Cost Target

Cost targets will be determined on a case-by-case basis.

Timeframe for Development

Phase 1: POC development: Q3-Q4 2025.
Phase 2: Commercial roll-out: To be determined on a case-by-case basis, target implementation by Q1 2026.

Potential Market/ Business Opportunity for the Product/ Solution

If the solution is successful, AECOM is willing to support further deployment across other locations worldwide. The solution could initially be scaled across various cities and regions, especially those facing high climate risks with high asset value (e.g. Singapore, Hong Kong). The solution also has potential in adjacent markets like utilities, transportation authorities, real estate developers, and insurance companies who are all interested in reducing climate-related risks.

As stated under performance criteria, we are open to exploring different business models together with potential solution providers and leverage our global network to create a win-win for everyone involved.

AECOM might be requesting for a period of exclusivity to bring the solution to AECOM clients first.

Resources

• Up to S$50,000 budget to support pilot development. POC development budget will be evaluated on a case-by-case basis depending on the quality and feasibility of the proposed solutions.

• Mentorship and support for solution development.
• Access to relevant datasets, lab facilities and pilot site(s)

• Up to S$20,000 grant support from EnterpriseSG on a matching basis

Other Considerations

We are looking for SMEs and startups with solutions that can be implemented in a relatively short-time frame (TRL of 5 and higher).
Intellectual Property (IP): For Background IP (BIP), both parties will retain their respective IPs bought into the project. In the event of new Foreground IP (FIP) creation, FIP ownership will be discussed on a case-by-case basis

Learn More

If you have an innovative AI solution, we invite you to submit an additional proposal under the "Open Category," where the corporates listed below are seeking cutting-edge technologies!

Proposals received may be reviewed by those companies, enhancing your chances of unlocking exciting business opportunities.

Please check the requirements listed for each statement. You may submit multiple proposals for different Open Category statements by creating a "NEW PARTICIPATION".

Generative AI Applications GOOGLE CLOUD

• Singapore-based AI startups only
• Pre-seed to Series A stage*
• Looking for AI solutions that are/have:
  • Leveraging machine learning and AI as core technology
  • Scalable and are tackling real-world challenges
  • High market potential with a significant total addressable market and a defensible, scalable business model

Shortlisted solutions** will be shared with Google for potential consideration to join Google for Startups Accelerator: AI First.

*Equity funding by an institutional investor or common web3 funding sources

**Terms and conditions apply.

Awarded innovator will receive support and collaboration opportunities.

Learn More

How might we find novel AI-based solutions to automate and improve our current sea surveillance process?

BACKGROUND OF THE PROBLEM

Safran Electronics and Defense operates in Singapore primarily in the aerospace and defence sector. Safran Electronics and Defense produces Optronics products that serve for Sea Surveillance. The purpose of our sea surveillance activities is to monitor existing ships and fleets as well as monitor incidents such as crashes of ships or planes in the sea to ensure operational teams undertake rapid action when required. The process of doing sea surveillance is often quite hectic especially since multiple individuals are involved and often the same individuals are involved in both sea surveillance as well as sea navigation. Today it takes about 5-10 seconds to do a cross-reference and verify an object.

For our sea surveillance activities, we are currently utilising a combination of (1) electro-optics and (2) helicopters and/or drones. The electro-optics allow us to look far into the distance, while the helicopters and/or drones provide a picture compilation. Based on these two methods, individuals then determine what objects might be present and whether or not further action should be undertaken.

In order to improve our sea surveillance process, Safran is therefore looking for novel AI-based solutions providers to build an AI-based system that can help us to detect the nature, size and number of objects based on variables such as movement, and warmth to give feedback to the operator in real-time. We are interested in these solutions to help us (1) reduce identification time, (2) reduce manpower cost, and (3) reduce error.

We have not tried any solutions thus far and are open to any solution that can solve our challenge utilising AI.

TECHNICAL REQUIREMENTS / PERFORMANCE CRITERIA

Technical Requirements:

1. Solutions should be able to identify an object within a 3-second (or less) time window.
2. Solutions should connect to existing surveillance systems to be able to do a real-time analysis.
3. The system should consider a ‘human-in-the-loop’ principle to always ensure maximum accuracy.
4. Ideally, the system should include dashboards with data insights such as: (a) type of objects detected, (b) occurrence of objects detected and related patterns etc.
5. Safran Electronics and Defense will make available the right amount and quality of data for the system to be trained upon. Data cannot be shared over the cloud and has to be kept in local servers for safety and security reasons.

Performance Criteria:

Safran Electronics and Defense is not able to provide specific performance criteria as these will be very solution-specific. Performance criteria will be generically evaluated by:

• Return on Investment (ROI) with a target ROI within 5 years
• We are looking for a 15% (or more) reduction of manpower through the system
• Payback period
• Quality of improvements
  

COST TARGET

Cost targets will be determined on a case-by-case basis dependent on the solution.

TIMEFRAME FOR DEVELOPMENT

Phase 1: POC development: Q1 2025.

Phase 2: Commercial roll-out: to be determined on a case-by-case basis, target implementation by Q3-Q4 2025.

POTENTIAL MARKET / BUSINESS OPPORTUNITY FOR THE PRODUCT/SOLUTION: