Commercial building AMR deployment presents challenges that differ from moving robots through a controlled warehouse. Shopping centres, office complexes and mixed-use developments contain changing routes, variable crowd levels, multiple operators and infrastructure shared by different robot fleets. Successful deployment therefore depends on coordinating the robots with building systems and peopleânot just selecting an AMR for a delivery task.
Technical guidelines developed by Singaporeâs Infocomm Media Development Authority (IMDA) Autonomous Delivery Workgroup address this operating environment. Their central themes are interoperability, real-time awareness of surroundings, reliable lift interaction and safety throughout the deployment.
Why commercial buildings need a different approach
Public-facing buildings are dynamic. A corridor may be narrowed by a temporary shop or promotional booth; a route may become obstructed by a crowd or spill; and customer orders can arrive unpredictably. Unlike many industrial settings, these environments are not necessarily managed by a single organization with control over every robot and task.
Several fleets may belong to different vendors, retailers or service providers. They must share pathways and, in many cases, interact with common infrastructure such as lifts and doors. Building operators and technology partners need ways to coordinate tasks and movement while accommodating changing conditions and varied robot capabilities.
The guidelines focus on earth-bound AMRs delivering within commercial property boundaries, including indoor and outdoor areas that are part of the property. They do not cover public-road operations or deployments in settings such as hospitals, hotels and schools.
Use interoperability to coordinate fleets and infrastructure
The Robotics Middleware Framework (RMF) provides a foundation for communication among different AMRs, building infrastructure and the services that support building operations. Its common messages and interfaces can help connect systems without requiring every vendor to use identical hardware or fleet-management software.
RMFâs modular functions support different parts of an operation:
- Dispatcher planning assigns tasks to available fleets according to defined criteria, such as resource availability or delivery efficiency.
- Traffic monitoring helps prevent conflicts between fleets through schedule negotiation, while allowing vendors to maintain their own fleet systems.
- Scheduling coordinates tasks and routes and receives updates about AMR status and location.
- Map and infrastructure management helps align maps and coordinate systems and manage shared assets such as lifts and doors.
The framework allows for different levels of integration. At higher levels, a system can monitor robots and assign tasks or routes. The guidelines identify medium-level complianceâmonitoring, task visibility and pause or resume capabilityâas the minimum acceptable level for commercial-building deployment. Limited integration, by contrast, may leave fleets dependent on manual scheduling and deconfliction.
Support real-time decisions with scene understanding
Standard task and traffic coordination does not, by itself, give an AMR enough information to respond to every unexpected event. Scene understanding extends RMF by processing image or video inputs and sharing relevant information with its core modules. The purpose is to help the system reassess a situation and support an appropriate operational response, rather than treating every deviation as a reason to abandon the task.
Inputs can come from cameras mounted on a robot or from fixed building cameras. Depending on the use case, modules can identify and locate objects, detect people around a lift, or identify temporary hazards. Their results can be published to RMF so that information from different robots or modules can contribute to coordination.
For example, if cones and tape mark a temporary hazard zone, scene-understanding software may identify the area and support a route update. Shared information can also help prevent other AMRs from planning paths through that zone. Near a lift, a safe-entry module can assess whether people are present and return a âgoâ or âno goâ result for the lift operation. These capabilities support decision-making; they do not remove the need for appropriate risk assessment and safety measures.
Design lift integration for normal and unusual conditions
Lift use is a critical part of multi-floor delivery. The guidelines describe a common node between RMF or a fleet manager and lift controllers. This arrangement can support communication with multiple lifts and robots, including commands from the robot-side system and status feedback from the lift.
Basic lift communication may not address every interruption. Extended messages can support time-out handling and convey information such as a liftâs current floor, door status, arrival at a request or destination floor, and operating mode. The system can use this feedback to determine what to do when the assigned lift becomes unavailable or an extraordinary operating condition occurs.
Lift selection also depends on physical suitability. Relevant considerations include car and doorway dimensions, rated capacity, landing access, floor levelling and the AMRâs footprint and gross weight. Operators should assess the robot and lift together, including how the AMR will enter, travel in and exit the car safely.
Make safety a system-level responsibility
AMR safety in a public-facing building involves more than the robotâs navigation software. The guidelines refer to established machinery and robot safety standards and add recommendations for issues associated with commercial-building operations. These include interaction with people, payload handling, lift use, environmental hazards and the behaviour of safety-related functions.
Deployment teams should consider hazards across the robot, payload, building infrastructure and operating environment. Safety functions addressed in the guidelines include safe stopping, speed and force control, stability, operational-space limits and environmental sensing. The appropriate performance targets depend on the function and risk assessment; the excerpted guidance does not justify applying a single target to every robot or site.
Operational planning matters as well. Routes, charging arrangements, access to infrastructure, fleet responsibilities and procedures for abnormal situations should be considered together. The guidelines also emphasize that different stakeholdersâincluding building owners, operators, AMR providers and infrastructure partnersâneed a coordinated approach.
Conclusion: plan the ecosystem, not just the robot
Commercial building AMR deployment is an integration project as much as a robotics project. Interoperable fleet coordination provides a basis for shared operations; scene understanding helps systems respond to changing surroundings; and lift interfaces connect mobile robots with essential building infrastructure. These capabilities must sit within a safety approach that accounts for people, payloads, machinery and unpredictable site conditions. For engineers and operators, the practical starting point is to assess the complete operating environment and define how every system and stakeholder will work together.
Access Official Technical PDF