A practical checklist for assessing hazards, people, controls and deployment readiness before introducing autonomous machinery.

Professional illustration of an autonomous excavator operating inside a controlled Singapore industrial yard, with a remote supervisor, marked exclusion zone and a safety-readiness checklist.

Autonomous heavy equipment is moving from concept discussions into controlled testing environments. For Singapore’s engineering contractors, facility managers, building owners and SMEs, the important question is not simply whether a machine can operate autonomously. It is whether the worksite, operating procedures, people and support systems are ready for safe and controlled use.

A useful local case study is the autonomous heavy construction equipment pilot launched by JTC and Kajima on 17 September 2026 at the Bulim Autonomous Yard. The pilot involves autonomous excavators and compactors, with conventional equipment retrofitted with autonomous capabilities and tested under real site conditions. Participating organisations include JTC, Kajima, the Public Sector Science and Technology Policy and Plans Office, HDB, Kok Tong Construction and Tractors Singapore.

The stated objectives include reducing reliance on skilled operators and keeping workers away from hazard zones. However, the technology is described as nascent and still undergoing testing, training and documentation. That distinction matters. A pilot is not the same as a production-ready deployment for every site.

What the pilot means for site operators

The pilot provides a practical reference point for businesses considering autonomous excavators, compactors or other mobile equipment. It also reinforces a broader technology-adoption principle: automation should be introduced with safeguards, human control and support for workers whose roles may change.

For a facility or construction-site operator, readiness should be assessed across six areas: the physical environment, risk controls, human responsibilities, communications, equipment support and deployment governance.

1. Map hazard zones before selecting technology

Start with the worksite rather than the machine. Identify areas where workers, visitors, vehicles and equipment may interact. This can include excavation edges, loading areas, temporary access routes, blind corners, reversing zones, overhead obstructions and areas with changing ground conditions.

Document when each hazard exists and who may enter the area. A static site plan may not be sufficient because routes, stockpiles, barriers and work fronts can change during a project.

Where supported by the equipment design and operating procedures, consider geofencing, controlled access and clearly defined exclusion zones. These should not be treated as substitutes for supervision or physical controls. The site team should understand what happens if a boundary is breached, a sensor becomes unavailable or the work area changes.

2. Define the human role clearly

Autonomous operation does not remove the need for accountable people. Before deployment, assign responsibility for starting and stopping the work, checking the operating area, authorising changes, responding to alarms and escalating incidents.

Do not rely on vague labels such as “remote operator” or “site supervisor”. Define the person responsible for each stage of the task. Clarify whether a worker is supervising one machine or several, what conditions require a return to manual control and who can suspend the operation.

A manual override or safe-stop function should be understood, accessible and tested as part of the operating procedure. The exact controls will depend on the equipment design, so operators should verify the manufacturer’s documented functions rather than assume that every autonomous machine behaves in the same way.

3. Test communications and remote supervision

Remote supervision depends on reliable communications between the machine, the control interface and the people on site. Review coverage across the full operating area, including locations where buildings, structures or equipment may affect connectivity.

Set out the response to lost communications. For example, the procedure may require the machine to stop, move to a defined safe state or await intervention, depending on the manufacturer’s design and the task risk. The response must be known before live operations begin.

Supervision also requires usable information. Alarms, machine status, location and task progress should be presented in a way that allows the responsible person to make a timely decision. If a control screen is difficult to interpret or produces too many alerts, the technology may create new operational risks rather than reduce them.

4. Build competency around new roles

Training should cover more than the software interface. Workers need to understand the machine’s operating limits, site boundaries, hazard controls, emergency actions, manual recovery and escalation process.

Competency records should reflect the actual role. A person conducting pre-start checks may need different training from someone supervising autonomous operations or performing technical troubleshooting. New roles may also require coordination between site operations, engineering support and the equipment supplier.

Singapore’s technology-adoption discussion has highlighted the importance of safeguards, human control and worker reskilling. In practice, this means involving workers early, explaining how responsibilities will change and providing a route to build the required skills. Treating automation as only a procurement decision can leave important operational gaps.

5. Strengthen inspection and maintenance procedures

Retrofitting conventional equipment with autonomous capabilities introduces additional systems that need to be checked and maintained. These may include sensors, control hardware, software interfaces, positioning components, communications equipment and safety-related functions, depending on the machine.

Establish a documented pre-start inspection process. Include the physical condition of the machine, the operating area, barriers, signage, communications status and any visible warning or fault indicators. Keep records of defects, repairs, software changes and return-to-service approvals.

Maintenance teams should know which work can be performed internally and which activities require the equipment supplier or qualified technical support. A machine should not return to autonomous operation simply because its mechanical fault has been repaired; the relevant autonomous and safety functions should also be checked before use.

6. Use staged pilot-to-production gates

A controlled pilot should have clear entry and exit criteria. Begin with a defined task, limited operating area, known equipment configuration and trained personnel. Monitor near misses, alarm events, communication interruptions, manual interventions, downtime and maintenance findings.

Before expanding the deployment, review whether the controls worked in actual conditions. Questions should include:

  • Were exclusion zones respected and understood?
  • Did workers know how to stop or escalate an unsafe condition?
  • Were alarms timely and actionable?
  • Did communication interruptions produce the intended safe response?
  • Were inspections and maintenance records complete?
  • Did the equipment perform consistently across changing site conditions?

Expansion should be approved only when the evidence supports it. A pilot that exposes limitations is still valuable if those findings are documented and used to improve the design, procedures and training.

How ISS can support readiness

For Singapore businesses, autonomous equipment may form part of a wider digital-transformation programme involving engineering operations, facility management, data collection and AI-enabled workflows. ISS can help organisations structure the discussion around their operating environment, current processes and practical automation requirements.

This may include reviewing operational workflows, identifying suitable pilot areas, documenting control points, improving reporting processes and considering how digital tools can support human decision-making. The objective is not to automate for its own sake. It is to introduce technology in a controlled, measurable and operationally useful way.

The JTC and Kajima pilot is a reminder that autonomous heavy equipment should be approached as a safety and change-management challenge as well as a technology project. Start with the hazard zones, define human control, verify the supporting systems and scale only when the evidence is strong.

Contact ISS to discuss engineering, facility management or AI automation requirements for your site or business.

Reference points