Cleaning Robots for Train and Metro Stations: 6 Key Selection Factors

Train and metro stations are in use from the first train to the last. Passengers and daily operations bring grit, moisture and litter into concourses, waiting halls, passages and platforms.

Two constraints shape floor cleaning: time and space. Large areas need cleaning around passenger peaks, while overnight access may need to be coordinated with other station activities.

Choosing cleaning robots therefore starts with the areas to be cleaned and the available operating windows. The machine and configuration should then match the station’s floors, cleaning tasks, access conditions and facilities.

Overview

• Suitable areas: Accessible floors in concourses, waiting halls, transfer passages and approved platform zones.
• Six selection factors: Cleaning areas, operating windows, safety boundaries, cleaning methods and utilities, team responsibilities, and network-wide deployment.
• Cleaning approach: Match sweeping, dust mopping or scrubbing to the soil, floor and robot configuration.
• Automation requirements: Assess charging, water handling and daily care alongside cleaning performance.
• Gausium experience: Published deployments span Shenzhen, Guangzhou, Madrid, São Paulo, Nagasaki, Chongqing and Network Rail stations.
• Key distinction: A successful deployment provides useful evidence, but its schedule and results are specific to that site.

1. Identify Which Station Areas the Robot Will Clean

Each area has different users, soil and access times. Planning area by area shows where a robot can undertake recurring floor work and where separate equipment or staff intervention is required.

At Chongqing East Railway Station, the published deployment covers the main waiting hall and transfer levels. At Network Rail stations, Gausium robots have been used in concourses, waiting areas and offices.

For shops and food outlets inside stations, see Gausium’s guide to cleaning robots for retail stores.

2. Fit Cleaning Into the Station’s Operating Windows

Published deployments show several ways to fit robot cleaning into a station’s day:

• Zone-by-zone daytime and nighttime cleaning: At Guangzhou Metro’s Caihong Bridge Station, the published schedule assigns the outer concourse to 23:36–01:06, transfer passages and the inner concourse to 01:10–05:30, platforms and the inner concourse to 09:30–15:00, and the outer concourse again to 15:30–17:00.
• Several sessions during service: On Shenzhen Metro Line 13, concourse robots run pre-dawn, morning, midday and evening sessions, excluding peak passenger periods.
• Platforms at night: On Shenzhen Metro Line 13, platform robots cover approximately 900 m² between 00:30 and 03:30.
• Overnight cleaning: At JR Kyushu’s Nagasaki Station, the robot cleans after the last train and before the first, dividing the 1,684 m² concourse into sections cleaned on alternate days.
• Off-peak operation with monitoring: On São Paulo Metro Line 5-Lilás, robots run mainly outside peak hours, with trained teams monitoring operations in real time.

These schedules describe individual deployments, not rules for other stations. Each operator needs to establish its own windows based on passenger movements, service patterns and access arrangements.

For further considerations in occupied spaces, see Which Autonomous Floor Scrubbers Can Clean High-Traffic Areas?.

3. Define Safety Boundaries Around Passengers and Platforms

Before routes and schedules are finalized, the station operator and implementation team should establish where the robot may work and which restrictions apply.

The site assessment should address:

• Platform edges: Exclusion boundaries defined by the operator.
• Tactile guidance paths: How the robot interacts with tactile paving and whether crossing is permitted.
• Boarding and circulation areas: Restrictions near train doors, fare gates, stairs and escalators.
• Emergency routes: Access that must remain unobstructed, including around parking and charging positions.
• Temporary changes: How closures, barriers or altered passenger routes affect assigned tasks.

Obstacle avoidance supports navigation around people and objects. It does not replace these operating boundaries or the station’s procedures.

Where daytime cleaning is proposed, testing should reflect the intended passenger conditions within the operator’s approved arrangements. Staff also need to know how to pause a task and when it may resume.

Gausium’s article on how commercial cleaning robots stay safe around people explains further considerations for occupied spaces.

4. Match Cleaning Methods and Water-Handling Arrangements

Dry methods such as sweeping and dust mopping collect suitable loose soil. Wet scrubbing washes compatible hard floors and recovers dirty water.

The floor and soil determine the required cleaning method. Water supply, drainage and docking arrangements then influence how the cleaning cycle is organized and which steps can be automated.

Two MTR-operated Shenzhen lines illustrate this distinction. Both use Gausium Omnie, but their deployed configurations differ.

The table describes these deployments, not the full capabilities of each brush configuration.

On Line 13, the installed docking setup and utility connections support automatic refill, drainage and charging. Line 4 uses dry dust mopping under its existing infrastructure conditions. This does not mean that every station without fixed connections must rule out robotic wet scrubbing.

Before deployment, assess:

• Water supply: Where refilling takes place and whether it is automated or staff-assisted.
• Drainage: Where recovered wastewater is discharged.
• Power and dock position: Charging arrangements that do not obstruct passenger or emergency routes.
• Return access: A usable route back to the dock from assigned cleaning areas.
• Staff assistance: Which replenishment and emptying tasks remain manual.

Omnie is designed for dynamic environments and uses AI to identify and clean dry and wet waste. Its disc-brush and roller-brush configurations should be selected for the proposed application.

For charging arrangements, see Which Cleaning Robots Can Charge Automatically?.

5. Assign Responsibilities to Station Staff and Cleaning Teams

Scheduled cleaning still depends on people at defined stages. Areas need preparation, equipment needs routine care, and alerts need a response.

The following table is an example of how responsibilities can be assigned. It should be adapted to the station’s organization.

Stairs, escalators, inaccessible areas and materials outside the robot’s approved scope remain separate tasks. Entrance matting and corners around fixtures should be assessed according to material, access and machine capability, rather than automatically assigned to either people or robots.

Routine maintenance and replenishment also need named owners across shifts. Automatic charging or water handling does not remove every equipment-care task.

For planning in occupied buildings, see 4 Machine Capabilities for Floor Cleaning in Occupied Public Buildings.

Train and Metro Stations Using Gausium Cleaning Robots

Gausium’s published station cases provide examples of different operating arrangements.

Historical deployments demonstrate the arrangements used in those projects. They should not be treated as evidence that every current model has the same configuration or performance.

6. Plan for Differences Across a Line or Network

Operators may assess several stations together. Stations on the same line can share design features, but each site still needs verification before a setup is reused.

Layout, floor materials, access boundaries and utility connections can differ. The two Shenzhen lines show how the same robot platform can be deployed differently according to those conditions.

A network-wide plan can define:

• Station groups: Similar layouts and facilities, with a proposed setup verified at each site.
• Schedules: Shared operating principles with station-specific exceptions.
• Support: A clear process for alerts, maintenance and service escalation.
• Results: Consistent methods for checking coverage, cleaning quality and resource use.

Alongside Omnie, Gausium Mira and Gausium Marvel are also options for subway station cleaning. Selection should follow the area, task and site conditions.

For larger facilities, see Cleaning Robots for Large Commercial Facilities. To assess robotic floor cleaning for a station or network, contact Gausium.

FAQ About Station Cleaning Robots

Q1: Can Cleaning Robots Work While Passengers Are in the Station?

Yes, within the areas and periods approved for the deployment. On Shenzhen Metro Line 13, concourse robots run several sessions during the operating day while excluding peak periods. Suitability should be verified under the station’s intended operating conditions.

Q2: Are Station Platforms Always Cleaned at Night?

No. Shenzhen Metro Line 13 uses an overnight platform-cleaning window, while Guangzhou’s Caihong Bridge Station includes daytime platform cleaning. The schedule follows each station’s service pattern and operating requirements.

Q3: Can One Robot Clean Both the Concourse and the Platform?

It depends on access and available cleaning windows. At Caihong Bridge Station, one robot serves concourses, platforms and a transfer passageway on a zone-based schedule. Shenzhen Metro Line 13 uses separate concourse and platform robots at each participating station.

Q4: Can Cleaning Robots Clean Escalators?

The floor-cleaning robots discussed here do not clean stairs or escalators. Those tasks require staff or dedicated equipment. Robot routes should cover accessible, compatible floor areas.

Q5: Who Handles a Robot Fault in a Station?

Responsibility should be defined before deployment. One arrangement is for the cleaning team to assess the alert and contact support, while station staff are informed if passenger areas or station operations are affected.

Sources

• Shenzhen Metro case study
• Guangzhou Metro case study
• Metro de Madrid case study
• São Paulo Metro case study
• Nagasaki Station case study
• Network Rail case study
• Chongqing East Railway Station case study