Home / Case studies / A large container port · Smart safety training centre
A container port ranked among the top in China by annual throughput moved high-risk operation training from the classroom into an assessable hands-on training space.
The port's operations include many high-risk scenarios — quay cranes, yard cranes, storage yards and gates — while traditional training relied mainly on classroom lectures and video learning.
The problem is not that we 'explained too little' but three other things: first, high-risk scenarios cannot be rehearsed repeatedly on the real work site; second, training effectiveness cannot be quantified, since attendance does not mean learning; third, results and training records are scattered across several systems, so when a supervisor inspects, no complete chain can be produced.
Safety training at ports has another particularity: staff turnover is high and new employees join in batches, so training must be completed at scale in a short time and leave verifiable records.
Delivered an integrated safety training centre that splits training into five stages — assessment → learning → practice → evaluation → management — each backed by a system.
In the assessment stage, AI personality assessment is used for job matching: people with a high risk appetite or slow emergency response are identified and routed before they start work.
The learning stage runs on an online academy that hosts courses and can push learning tasks by role.
The practical stage covers eight or more high-risk scenarios with VR and physical equipment, and operations are logged.
In the evaluation stage, the comprehensive assessment control system consolidates data from every stage and builds competency profiles for individuals and teams.
In the management stage all records are consolidated and transcripts are attested on-chain, forming a tamper-proof training archive.
The table below lists this project's technical elements so clients can verify them independently.
Project stories without constraints are fiction. This section sets out four constraints: time, site, staffing and compliance.
These constraints directly shape the solution; every problem under 'Key challenges' below is forced out by one of them.
It is not “we worked hard”; it is “what the problem looked like, how we solved it, and what it cost”.
This section is for peers. If you are doing something similar, these points will save you a few detours.
For any system aimed at front-line staff who cannot leave their posts, start with units of 10 minutes or less before considering a full course.
Generic scenarios are good for demos, not for assessment. The cost of rebuilding a scenario has to be budgeted up front.
It satisfies the tamper-proofing requirement without triggering compliance issues over data leaving the network. The same pattern applies to any case where you must prove something but the data cannot leave.
Talking only about successes is unconvincing. This section covers what we got wrong or did too late.
Doing it again, we would spend a month integrating backend data before starting the front end. We did the interface first and wired data afterwards, which forced a mid-project rework: three systems used different field definitions and three different staff ID schemes, ending in a full mapping pass.
The effort of rebuilding VR scenes was badly underestimated. We quoted it as swapping textures; in practice it was a rebuild from the real site. The next quote will show this as a separate line item.
Offline capability came too late. Signal is poor in parts of the port, and the app's offline mode was only added in the third month. Any system used on site should be offline-capable from the first release.