Innovation Challenges

Robotic Cable Pulling in Conduit Systems

Conduit pipes are commonly used to contain electrical and signal cables within buildings. The length of cables used in each project varies widely, ranging from hundreds of metres in smaller residential developments to hundreds of kilometres in larger buildings such as hospitals. 

What We Are Looking For

The solution should:

• Be able to pull the pilot cable through existing conduit systems with complex bends; and
• Reduce no of manpower involved in the cable installation process.

Learn More

Robotic Concrete Slump Testing and Cube Making

The slump test is a basic quality control test that determines the workability of fresh concrete. Cube making is a method to determine the compressive strength of hardened concrete. Both are essential to ensure the high quality of concrete used in construction projects. 

The current practices in both are time-consuming and labour-intensive. Further, the manual nature of these tests means they are prone to human error and can be affected by factors, such as the skill and concentration level of the operator. These inconsistencies can result in problems, as it is critical for testing methods and measurements to exactly align with the EN standard. Additionally, it is becoming increasingly challenging and expensive to hire skilled workers to perform these tests.

What We Are Looking For

The solution should: 

• Complete the process within a reasonable time;
• Be portable and thus easy to transport to different sites;
• Require minimal training to use, making it accessible to a wide range of users;
• Require minimal manual intervention;
• Result in cost savings in terms of time and manpower;
• Be able to operate in environments with: some text
  • Shaded temperature of 20-40°C
  • Humidity of 50%-90%
  • Minor levels of dust
  • Minor levels of moisture
• Be energy-efficient; and
• Produce minimal waste and emissions.
  

Specific for slump testing:

• Perform concrete slump testing accurately according to EN 12350-2 and EN 12350-8 standards; and
• Auto-measure the slump and slump flow of fresh concrete with minimal error.
  

Specific for cube making:

• Perform cube making accurately according to the EN 12390‐2 standard; and
• Automatically shape and compact concrete cubes with minimal error.

Learn More

Robotic Painting of High-Rise Building Façades

Painting works on external walls of buildings are usually carried out using rollers and brushes. To work on high-rise buildings,workers are hoisted on suspended scaffold gondola systems.  Despite stringent safety protocols, the inherent risks of working at considerable heights. If the safety procedure is not followed properly, or if an unexpected equipment malfunction occurs, a serious accident may result. 

Additionally, it is becoming increasingly difficult to recruit workers, resulting in a manpower shortage. It is thus becoming necessary to look into ways to reduce reliance on manual labour for building painting works. 

What We Are Looking For

The solution should:

• Negate the need to use gondolas, boom lifts, and other work-at-height equipment;
• Be able to operate unpiloted during the painting process;
• Be compatible with currently available painting materials and technologies;
• Be compatible with building mapping standards (for example, DJI Mapping and BIM) to enable automated painting;
• Enable paint to be applied to at least 80% of external walls; 
• Enable paint to be applied without contaminating the surrounding environment (that is, no overspray); and
• Achieve improvement in productivity by 50% compared to conventional methods.

Good to have:

• Be able to apply paint to hard-to-reach surfaces or corners.

Automated Quality Inspections for Building Interiors

CDL’s residential projects are well regarded due to their high quality. This can be attributed to strict and robust quality assurance (QA) and quality control (QC) checks conducted before the handover of each project. These processes are designed in-house, tapping on a wealth of experience and information from past projects to ensure that the execution of the inspections is continually optimised. 

What We Are Looking For

The solution should:

1. Using a high resolution 360° camera or similar device, detect and identify visual defects that are visible to the naked eye (including, cracks, chips, scratches, stains, peeling paint, rust etc.)
2. Accurately identify and categorise at least 70% of the detected defects;
3. Complete the inspection of a single residential unit between 10 and 15 minutes (compared to the approximate 45 minutes taken by by 2 human inspectors);
4. Be integrated with a digital platform, which can be pre-existing or specially developed to support your solution;
5. Automate the process of flattening the images;
6. Generate a full defect report for a single residential unit within 48 hours;
7. Build up a database of defect images to help train the AI model to achieve higher accuracy; and
8. Provide suggestions to optimise the QA/QC inspection process.

Learn More 

Automated BIM Database Mapping

Some construction projects require a massive amount of information to be embedded into BIM models. To prepare BIM models which fulfil the level of information required, there are many datasets and essential data attributes (for example, floors, walls, columns, windows, and doors) that need to be tagged to Revit families..

The process of tagging these essential data attributes is done manually, with additional manpower often deployed for this. Tagging is a skilled task that not all staff at consultancy services companies like CPG Consultants, are familiar with, thus presenting a significant operational challenge.

What We Are Looking For

The solution should: 

• Be able to integrate seamlessly with Revit and other common BIM software where applicable;
• Be able to integrate seamlessly with multiple mapping databases (for example, IFC-SG, MCR, and GWP);
• Enable all Revit families (for example, floors, walls, columns, doors, and windows) to be automatically and accurately tagged with essential data and parameter values from CPG databases;
• Automatically check for the accuracy of parameter values;
• Automatically create system families and classify objects accurately into each family;
• Establish and streamline workflows to tag typical rooms modelled in Revit, including key Revit families (for example, floors, walls, columns, doors, windows, equipment, and furniture). This should apply to other common BIM software where applicable.

Additionally, the solution provider should work towards measuring and quantifying the time savings associated with their solution, in comparison to conventional manual methods.

Learn More 

Automated Cost Estimation and Compliance Checks

Despite advances in digitalisation in the built environment sector, many interior design and build projects still rely on manual quantity take-off (QTO) and compliance checks. This process involves many different stakeholders: architects, engineers, quantity surveyors, and consultants. To add to the complexity, there is no standard content or data format amongst them. 

What We Are Looking For 

The solution should:

• Detect and identify at least 80% of typical objects (furniture, walls, doors, and rooms);
• Incorporate a library of objects that is large enough to support the AI applications for QTO and compliance checks throughout design, fabrication, construction, asset delivery, and facility management (AI library development may be phased and prioritised depending on the project development needs);
• Link common open data (e.g. BuildingSMART Institute’s IFC Schema, Singapore’s IFC-SG) and BIM objects with the Building and Construction Authority’s (BCA) Model Content Requirements (MCR) parameters and its project delivery and compliance checks;
• As a start, incorporate IFC-SG schema to auto-classify recognised objects based on VML, with primary information to be attached to individual identified objects and trades;
• Categorise these identified objects with at least 70% accuracy;
• Provide a visual-based platform for various professional disciplines to work with;
• Provide a summary of QTO data in formats tailored to various disciplines, allowing professionals to bypass the data organisation phase (especially for large projects) and proceed directly to data review and adjustment. For example:some text
  • Space requirement matching and optimisation for interior designers
  • Headcount and occupancy-related computations on fire services/ACMV for engineers
  • Budget optimisation and value engineering processes for quantity surveyors
• Enable the data to be viewed on a live visual platform with user-centric UI and UX.

Learn More

Automated Project Scheduling

When developing a detailed schedule for a new construction project, the planner needs to manually input the details of every task to be performed. A typical construction project involves thousands of tasks, requiring planners to handle and provide large amounts of information, from duration and sequence, to resource type and quantity.

What We Are Looking For

The solution should:

• Be able to read and edit Oracle Primavera P6 (.xer format) and Microsoft Project (.mpp format) files;
• Be able to handle multiple files from multiple projects (can go up to several hundred .xer and .mpp files) at any one time;
• Collect data on estimated and actual time taken for completion of tasks;
• Maintain an updated database of productivity figures for various projects (see resources for more details);
• Provide insights and recommendations to improve the accuracy of project schedules;
• Input productivity figures to modify new project schedules in .xer and .mpp formats;
• Allow results (productivity range, evolution over time, based on resource number,etc) to be exported in Word and Excel formats;
• Allow users to customise and select categories, items, and task groups for each project (for example, “group by civil, electrical, MEP disciplines” or “group by detailed tasks: rebar installation, formwork installation, finishing”).

Autonomous Inspection of Building Façades

Periodic façade inspection (PFI) is mandated by BCA for all buildings older than 20 years and taller than 13 metres. Approximately 30,000 buildings in Singapore are to be inspected every seven years – or around 4,000 buildings a year. This regulation may in future be extended to newer buildings, further increasing the number of buildings that have to undergo routine PFI.

What We Are Looking For

The solution should: 

• Allow a human operator to view a video stream or take photographs to view the condition of the cladding framing, bracket, and anchors, and examine closer for issues;
• Be able to move autonomously or via remote control;
• Be able to move on both horizontal and vertical surfaces;
• Be able to access narrow cladding cavities, which generally range from 50mm to 150mm; and
• Be able to navigate obstacles within the cladding cavity, for example, rainwater down pipes, brackets, runners, lightning protection tapes, etc.

Learn More

Autonomous Inspection of HDB Corridors

As the Managing Agent (MA) for various Town Councils in Singapore, EM Services is responsible for carrying out monthly routine inspections of the corridors and stairways for all Housing and Development Board (HDB) blocks under their jurisdiction.

What We Are Looking For

• Enable real-time connectivity and video streaming to a centralised Command Centre;
• Preferably enable real-time video analytics and identification of defects such as hazards, faulty lights, illegal placement of objects, etc.

The solution should:

• Perform the role of a PO, negating the need for staff to physically be present at the inspection site;
• Be able to forward or backward autonomously in the common corridors, and manoeuvre spaces that are narrower than 1.2 metres in width (Note: While standard HDB common corridors are 1.2 metres wide, residents often place objects such as flower pots and shoe racks in the space);
• Take lifts on its own;
• Be equipped with warning lights and voice annunciations to make its presence known to approaching human beings
• Give way to human beings, and in situations where the robot cannot resolve “right of way” conflicts or becomes stuck due to obstructions, there should be an option to remotely pilot the robot from the Command Centre;
• Operate quietly as it will be deployed between 11pm and 5pm, when there is less human traffic;
• Have a battery life of at least 6 hours;
• Capture images/point cloud data of the corridor and generate a 3D model of the corridor;
• Identify and classify defects, hazards, and obstructions, and pinpoint defect locations on a 3D digital twin of the residential block;

With the increasing deployment of service robots in housing estates, there is an emerging need to manage different robot types and makes, and thus a platform with a common interoperability framework (such as TR93) to manage these robots.

Learn More

Autonomous Inspection of Lift Shafts

The process of installing lifts in a building begins with creating shop drawings based on architectural plans. These typically 2D drawings depict the lift layout through plan and elevation views. Once the building is topped off, EM Services begins the installation process. The lift shaft is usually empty. The lift engineers have limited capability to inspect it until a fixed (scaffold) or moving (example, gondola) working platform is set up.

What We Are Looking For

The solution should:

• Allow for the dimension and location of the rail brackets to be determined without having to install plumb lines;
• Perform a contactless scan of the entire lift shaft to generate a 3D CAD model with an accuracy of +/- 1mm, allowing for inspection to be done off-site by a Specialist Professional Engineer (SPE);
• Provide 3D video and/or photos with sufficient resolution for offline analysis, in conjunction with the 3D scanning 
• Enable the use of BIM for lift installations: with the 3D model of the as-built shaft, the 3D model of the lift assemblies could be fitted into the lift shaft to check for clashes and clearances;
• Map a 50-storey shaft (approximately 150 metres) in 60 minutes or less;
• Supplement the annual mandatory inspection with a scan of the shaft and analyse dimensional variances from earlier scanned models to any changes in the shaft, such as guide rail misalignment; and
• Accumulate a library of detected defects, which can then be used to enhance the auto-detection.

Learn More

Autonomous Inspection of Roads for Subsidence

Road subsidence and collapse is an issue in large urban cities, due to the intense and large-scale development of underground spaces under which these roads lie. When excavation is carried out to prepare for the installation of new infrastructure, it can result in displacement of the geological structure of the underground spaces. For example, the construction of MRT tunnels can add complexity to the original underground makeup, resulting in a loss of structural integrity. Additionally, rapid urban development results in an increased prevalence of issues like leaks in underground pipes, incomplete backfilling of deep excavation, and rainwater scouring of road structures. As a result, the risk of road subsidence increases.

Furthermore, in large coastal cities like Singapore, geological conditions such as decline in groundwater level and the soil structures can result in instability of foundations, making them vulnerable to traffic loads and other environmental factors. This further raises the risk of road subsidence, necessitating regular monitoring of roads to detect and mitigate road subsidence hazards as early as possible. 

Currently, handheld ground-penetrating radar devices are used for inspection, with the aim of detecting areas with high levels of road collapse hazards. However, with so much ground to cover, conducting road inspections manually is an expensive, tedious and time-consuming process. Therefore, Asia Labs sees an opportunity to automate the process of road inspection to ensure that more area can be covered in a significantly shorter amount of time.

What We Are Looking For

The solution should:

GPR equipment

• Be lightweight enough to be mounted on a small vehicle such as a car or pick-up;
• Be detachable;
• Be multi-band with range of frequency of 50 t0 2000 MHz in order to detect targets ranging from 1cm to 50cm; and
• Have a roadbed detection depth greater than 2.5m and resolution less than 10cm.
• Able to detect the following hidden road defects which include:some text
  • Rupture and leakage of underground pipe network;
  • Incomplete backfilling of deep excavation;
  • Rainwater scouring of road structures;
  • Decline in groundwater level; and
  • Soft soil foundation structure

Automated radar data interpretation software

• Detect abnormal images picked up by the GPR with at least 90% accuracy;
• Identify infrastructure diseases from abnormal images with at least 80% accuracy;
• Be able to process data at a rate of at least 100 km/day; and
• Have a software interface that is compatible with mainstream hardware devices such as the GPR and other equipment used for the solution.

Learn More

Long-term Monitoring of Deep Foundations

Deep foundations are used in many types of civil infrastructures, such as high-rise buildings, bridges, railway lines, and tunnels. The design and installation of these deep foundations are crucial to ensure that there is sound support and ground stability throughout the lifespan of the structures built above. In Singapore, deep foundations, such as cast in-situ bored piles, are common due to a combination of the type of soil strata and the prevalence of high-rise buildings and other urban infrastructure.

What We Are Looking For

The solution should:

• Eliminate the need for manual inspections of foundation piles and hence, reduce manpower;
• Enable automated data capture of deep foundation operating performance over a long period (25+ years);
• Provide reliable and accurate data with resolution down to single digit microstrain in real-time; 
• Use a wireless communication protocol that is robust enough for on-site deployment with limited signal loss or data drop;
• Provide a dashboard that allows users online access to updated data; and
• Send alerts to users if measurements exceed the normal range.

Learn More

Edge AI Video Analytics for Monitoring Site Safety

Video analytics systems for construction have been quite effective in monitoring site operations for safety, productivity, or security and providing real-time actionable insights for better decision-making.

However, deploying these solutions at construction sites presents significant challenges, primarily stemming from the intricate networking required to transmit CCTV streams reliably and in a cost-effective manner either through 4G/5G, wireless bridges, or mesh networks. 

What We Are Looking For

The solution should encompass real-time video streaming for remote site oversight, facilitated by a robust command and control system. Furthermore, it must integrate AI-powered analytics directly at the edge device, enabling proactive detection and response to safety breaches.

The solution should: 

• Encompass a self-contained unit furnished with essential components, including cameras and supporting infrastructure (for example, poles, waterproof enclosures, and power regulators) for seamless operation.
• Include provisions for cooling the edge devices.
• Seamlessly integrate with any IP camera, regardless of the brand, to facilitate the execution of custom safety video analytics use cases.
• Enable the deployment of custom AI models and safety use cases directly onto the edge AI device.
• On the chosen edge device, demonstrate the ability to process a minimum of 2 camera streams simultaneously in real-time, with each stream supporting at least 3 distinct AI-powered safety monitoring use cases.
• Enable the edge AI functionality to maintain continuous analysis even during periods of low to zero internet connectivity, transmitting alert information once the signal is restored.
• Enable all generated alerts to be accessible on both a mobile platform and a web-based dashboard, providing comprehensive visibility into alert information; and
• Develop APIs to facilitate seamless transmission of alerts to any external third-party system or Exyte's proprietary systems.

Please download a document (PDF) containing detailed requirements and use cases for the solution.

Learn More

3D Digital Twin Model of Building Façade

Undertaking work on building facades can be a challenging task, especially when it comes to coordinating different specialists and the asset owner. It is even more difficult when dealing with buildings that have complex geometry, such as curved facades, recessed areas, and numerous returns. In such cases, a large set of drawings is usually required to document the geometry, making the process even more cumbersome.

Moreover, when undertaking Addition and Alteration (A&A) works, it is crucial for contractors to know where the works are located and potential access restrictions on the building. This can be a time-consuming and complicated process, especially when relying on spreadsheets and written reports.

What We Are Looking For

The solution should:

• Create a graphical representation of the building that enables the identification of specific facade elements for observation tagging purposes and facilitates a thorough comparison between the model and the pre-existing structure;
• Effectively distinguish between distinct types of façade elements, such as glazing, cladding, louvres, signages etc.;
• Use recording and retrieval processes, resulting in considerable time savings compared to the traditional method of using 2D drawings;
• Enable real-time monitoring of results and images to ensure transparency;
• Ideally implement a web-based model that is accessible to all vendors, including clients, contractors, engineers, and service providers, through an API, thus ensuring compatibility with various systems and enhancing accessibility for all users; 
• Provide a platform for comprehensive tracking of historical data, including replacements, rectifications, window cleaning, BMU servicing, anchor point testing, defects, etc.; and
• Be adaptable to any forthcoming photogrammetry models, enabling the user to toggle them on or off.

Learn More

Automated Tracking of Earthworks and ERSS

The tracking of manpower and equipment productivity, and project progress in construction sites is done manually. The main contractor is required to compile productivity data from their subcontractors and turn them into periodic reports. On top of this being time consuming, the current practice results in missing or unreliable data.

In civil projects, early construction is focused on earthworks and installing the Earth Retaining Stabilising Structure (ERSS). At this stage, utilities, equipment, vehicles, and even CCTVs are often moved around the site. The frequency and extent of such movements makes it challenging to track productivity manually.

What We Are Looking For

The solution should:

• Be able to read data from site CCTVs; 
• Have 90% accuracy in object and activity detection (see resources for full list of activities to be tracked);
• Be able to detect unexpected downtimes and provide automatically calibrated productivity insights;  
• Provide a platform, including a customisable dashboard, that aggregates the site activity data into a dashboard with useful insights 
• Enable automated PDF export of regular (option to choose daily, weekly or monthly) progress and productivity reports;
• Be securely hosted with access enabled only by user logins.

Good to have: 

• Direction integration with Microsoft Project or Primaverato update percentage-completion progress of each activity; 
• Integration with and hosting the dashboard on the common data environment (Fulcrum);
• Enable 3D model visualisation of earthworks progress by photogrammetry;
• Enable BIM model visualisation of piling and D-wall progress, minimally supporting Autodesk Revit and International Foundation Class (IFC)

Learn More

Integrated Automated Reality Capture and Progress Tracking

Progress monitoring in construction sites is a highly manual process. With the large number of stakeholders involved in a project – developers, sub-contractors, suppliers, and engineers – challenges in collecting data and tracking processes not only result in unnecessary delays and costs, but also difficulties tracking errors in works. That can in turn lead to deviations from planned designs.

In greenfield construction, the project design is developed using a Building Information Modelling (BIM) model, which sub-contractors adhere to when carrying out works. While 3D LiDAR scanners can be used to facilitate progress tracking, the post-processing and report generation stage remains a manual process. 

What We Are Looking For

The integrated solution should encompass data capture, point cloud stitching, post-processing, deviation identification, clash prediction, and report generation. The solution should: 

• Seamlessly integrate with various scanning robots such as wheeled or legged robot (e.g., Boston Dynamic, AgileX Scout 2.0) ensuring compatibility and adaptability across different robotic platforms.
• Enable robots to autonomously navigate in a factory environment using predefined way-points;
• Enable efficient and accurate automated post-processing of 3D point cloud data obtained from the scans such as auto registration and auto denoising (people removal, temporary works removal if have reference BIM model, outliers noise interference) of point cloud data;

For greenfield projects:

• Calculate deviation differences between planned and actual elements and perform a thorough analysis of variations between planned and actual elements within a BIM mesh and point cloud, ensuring a minimum 5-millimetre precision. Include calculation of variances, incorporate a difference feature for each point, and deliver a comprehensive deviation analysis in a PDF report; and 
•  Identify potential clashes based on current deviations, considering the original clash-free model, in other words highlighting how deviated as-built elements will impact elements that have yet to be built.

For brownfield projects:

• Automatically generate a 3D BIM model based on the collected point cloud data, with the option to generate 2D floor plans based on this 3D BIM model;
• Automatically identify general components, such as fire extinguishers and other equipment, based on point cloud data; and
• Precisely measure the distances between individual components and represent them accurately on the generated floor plans.

Please download a document (PDF) containing details of detailed requirements of data capture and point cloud stitching, post-processing and deviation identification, clash prediction, and report generation.

‍Learn More