Konecranes

  • ( Crane training )
  • ( Networked )
  • ( Unity )

VR training for overhead crane operation, built in Unity for Quest. The trainee works the trolley, bridge and hook joysticks, moves a load through a scenario, and is judged on how the load behaves rather than on where the crane ends up. Sessions can be networked, and the whole thing can be driven by real control hardware over Bluetooth.

The sway is the lesson

Crane operation is mostly about what the load does after you stop moving. So the load simulation runs weight under gravity, pendulum sway, a spring connection to the hook and air resistance. A trainer where the load hangs still teaches nothing worth knowing.

Operator and headset views side by side: the load on its pallet, hook above it, control panel bottom right.

Joysticks with mechanical character

Each of the three joysticks accelerates and decelerates rather than moving in binary, and fires events when it reaches its constraint limits. A hook centring filter enforces lifting from directly above the load, which turns a safety rule into a control behaviour instead of a warning message.

Teaching the grab before the task

Dummy joysticks animate on their own to show what to reach for, and disable themselves the moment the real ones are grabbed. It solves first-contact confusion without a line of instruction text.

Assessment made of geometry

Rather than a bespoke script per exercise, we compose scenarios from small triggers: is the load within tolerance of a target height, is its tilt within leeway of a reference, is the horizontal offset inside a threshold, has a joystick hit a limit. New exercises come from recombining conditions rather than from new code.

A physical control desk driving the simulation

An ESP32 talks to the application over Bluetooth Low Energy, so the same scenarios can run against real hardware controls instead of tracked hands. That matters when the training has to transfer to a cab.

The physical pendant in the trainee's hands, driving the crane in the headset.

One authority in a networked session

Multiplayer runs on Fusion, and the rule throughout is that the host owns movement. Followers, position animations and hand-ghost switching run host-only with clients following, and networked objects disable themselves on spawn until the host says otherwise.