Autonomous Cleaning Robot for Construction Site Maintenance

Automation & Robotics for Construction

Challenge Statement Owner

Penta-Ocean is an industry-leading Japanese construction firm with widespread business presence internationally. Since the establishment of the International Business Unit in Singapore, Penta-Ocean has been delivering land reclamations for Marina Bay, Tuas, and Changi Airport, before expanding into civil engineering and building works.

Background

Across a construction project lifecycle, multiple stages demand regular cleaning. These include daily maintenance of access ways, cleaning before the application of architectural finishes, and final pre-handover cleaning. Current practices rely heavily on manpower, which introduces variability in performance, safety risks, and rising labour costs amid workforce shortages.

Construction site cleaning is labour-intensive, reactive, and inconsistent. Workers are assigned to manually sweep and wash dusty slabs, remove pooled water, and clear muddy paths. In outdoor conditions, mud accumulates on steel plates and slabs used for transporting materials, while in indoor basement zones, poor drainage causes water ponding. This leads to safety risks including slips and mosquito breeding.

Previous attempts to automate cleaning, such as using mechanical sweepers and off-the-shelf vacuum robots, have failed due to poor suction, mobility limitations, and an inability to operate effectively in harsh environments with low lighting or uneven terrain.

Penta-Ocean sees an opportunity for a site-capable autonomous cleaning robot that can reduce dependence on manual work while delivering consistent, high-quality cleaning outcomes in dynamic site conditions.

‍

The Challenge

How might we develop an automated construction site cleaning to reduce manual labour and maintain consistent site hygiene?

Requirements

Functional Requirements

Perform multiple cleaning modes including sweeping, brushing, vacuuming, and pressure washing across different site surfaces
Navigate common construction materials such as steel plates, concrete slabs, and rough or uneven terrain
Detect and avoid obstacles, including kerbs, holes, and pipe penetrations ≥100mm
Operate effectively in wet, dusty, and low-light conditions, including in semi-enclosed basement areas
Generate regular cleaning progress updates, including system status alerts and cleaning completion reports

Technical Requirements

If an autonomous robot solution:

Movement and navigation capabilities and features:
- Can move autonomously, but with sensors to avoid obstacles in direct path; avoid going into restricted areas that are fenced up; and avoid falling through openings, in case it manages to go past fenced areas
- Auto-homing when battery is low and resume cleaning from last cleaned point

Cleaning capabilities and features:
- Jet or pressure washing capability for mud and laitance removal
- Brushing and sweeping for dust and debris
- Vacuuming for finer particulate matter
- Self-emptying tank or waste container for extended autonomous operation
  ‍
Smart capabilities and features:
- Computer vision to know when jet/water washing is required at localised or general areas
- Remote route update – If an autonomous route cannot be followed due to site utilisation changes, to notify the user that an updated route is required
- Remote access to robot status, system health, and cleaning task progress
- Real-time dashboard for monitoring cleaning coverage and completion
- Ability to generate and export cleaning performance reports

Additional Preference:

Modular configuration for outdoor and indoor functions (if dual use)

‍Expected Outcomes

Reduction of manpower – man-hours reduction up to 80% expected – for daily cleaning at the work siten speed by at least 30%
Improved site safety and hygiene standards
Reduction of mosquito breeding risks (still water ponding) and health hazards

Deployment Environment and Constraints

The robot will need to function across semi-finished structures. The environment is often wet, dusty, poorly lit, and at the same time, constantly changing. Sites include outdoor areas where steel plates collect mud, and indoor basements where stagnant water and dust pose health and safety hazards. Connectivity may be limited, particularly underground. The system must withstand construction debris, rain, and floor level differences. It must also handle a dynamic obstacle landscape that includes both fixed features (kerbs and upstand pipes) and daily changes from new walls or temporary material placements.

Proof-of-concept (POC)/Pilot Support

Penta-Ocean will facilitate the proof-of-concept by providing access to an active project site(s) that represent varied and realistic cleaning conditions:

Elective Care Centre (ECC), National Dental Centre Singapore (NDCS) and Rapid Transit System (RTS) Link Contract T232 projects – all in the finishing stage, and thus ideal for testing dry surface cleaning ahead of tiling and client handover
Air Traffic Control Centre (ATCC) project – a top-down construction site offering basement environments with muddy surfaces and water ponding, and thus suitable for testing the robot’s wet cleaning capabilities

The POC will be structured around clearly defined performance metrics, including cleaning coverage per session, system uptime, obstacle detection and avoidance accuracy, and overall operational reliability. Testing will simulate routine site conditions to assess the robot’s effectiveness in replacing manual cleaning processes.

Penta-Ocean will provide:

Access to designated cleaning zones in a live construction site(s)
Historical and live cleaning data, including cleaning frequencies
Input and evaluation from subject-matter experts in safety, engineering, and digital delivery
Support in co-defining success criteria, KPIs, and cleaning quality benchmarks
Dedicated coordination to facilitate on-site logistics, safety compliance, and integration with daily site operations

‍Commercialisation and Scaling

Upon successful pilot, Penta-Ocean will adopt the solution across its portfolio of major construction projects. The robot will be deployed through a leasing model priced between S$2,500 and S$6,000 per month depending on functionality. The company aims to set a new industry benchmark for automated cleaning on construction sites, replacing inconsistent manual cleaning with predictable, data-driven processes. The solution could also be offered to other contractors and facility managers across infrastructure, industrial, and commercial projects.