Geo reference

​

What this is for

You have a scan of a real place — a factory, a yard, a warehouse — and a machine moving around inside it that reports GPS. The machine knows where it is on Earth. Your scan does not. A geo reference tells the environment where on Earth its origin sits, which lets the two be reconciled: the robot appears in the right aisle, in real time, in your point cloud. Live mode only. GPS is not simulated. In Simulate and Edit a GPS-positioned twin sits at its saved position.

​

Set the geo reference

In Edit mode, right panel → Geo Reference: Save, and the reference applies to every GPS-positioned twin in the environment — they move to the corrected coordinates straight away, without waiting for their next fix.

​

Show the map under your scene

Once the environment has a geo reference, you can put real-world imagery on the ground beneath it. View → GPS map, or the globe button in the toolbar. The map is turned to match your Heading, so scene north is real north. Twins, waypoints and areas sit on top of it, which makes it the quickest way to confirm a geo reference is actually right: pick a feature you recognise — a building edge, a road, a fence line — and check the scene lines up with it. If the whole scene looks rotated against the imagery, the Heading is wrong. It opens straight down, as a plan view. That is the view to place and check points in: looking at the map from an angle puts the sides of buildings between you and the ground, and a point you click can land metres from where you meant. Pan and zoom as much as you like — the view stays flat until you ask for a tilted one. Available in Edit, Simulate and Live. Unlike GPS positioning, the map needs only the geo reference, not a live fix, so it works in all three. There is no size or extent to set: pan and zoom the map and it loads the ground you are looking at. Without a geo reference the toggle stays disabled — there is nowhere on Earth to put the imagery yet. Set the reference above and it becomes available. If the environment was cloned, confirm and re-save the reference first.

​

Sat, Map, 3D

The switch at the top right of the viewer picks the view: The scene tilts with the map, so in 3D your twins stand up on the ground next to real buildings instead of reading as marks on a floor plan. 3D uses the road map, not satellite — and switching to it changes the imagery for you. Buildings are geometry in Google’s drawn map; satellite is photography, so tilting it gives you the same photo at an angle rather than a city you can see the height of. Both flat views are the honest ones for placing and checking a point: straight down, one screen pixel to one spot on the ground. Coming back to 3D returns to the tilt you last had, so flattening to check something does not throw away a view you set up by hand. View → Top and View → Isometric move between flat and tilted too. The compass, rotate and pitch buttons underneath handle orientation. While the map is flat, dragging cannot tilt it — press 3D or a pitch button to leave the plan view, so an accidental drag never quietly angles a view you are picking points in. Each of those buttons works two ways. A click steps: 30° of rotation, 15° of pitch. Holding the button — or holding its keys — turns the map continuously until you let go, which is how you line the view up with something rather than counting presses. A hold starts slow and eases up to full speed over about half a second, so a tap nudges and a hold sweeps. Letting go of either modifier stops the movement, as does releasing the arrow. The keys stand down while a robot is under keyboard control, so a keypress never both drives a robot and moves the camera. The buttons keep working either way. Google’s imagery runs out before the view does: satellite stops at about 5 cm per screen pixel, the road map one level closer, and a robot tens of centimetres across is only a few pixels there. So zoom carries on past that point — twins, waypoints and areas keep being redrawn sharp while the imagery beneath them is enlarged and goes soft, up to 16× closer than Google’s own limit. The microscope button at the bottom of the control cluster switches that on and off, and shows how far past the limit you currently are. Turn it off when the imagery is what you are looking at — lining the environment up against a building edge, or taking a screenshot — and the view stops where Google’s sharp imagery stops. Leave it on to inspect a twin. How much 3D you get is Google’s data, not a setting: buildings show up in covered areas at close zoom, and a rural site may stay flat however far you tilt. Google also limits the tilt as you zoom out, so zoom in before judging it.

​

Type a position as latitude and longitude

Anything you can place — a twin, a waypoint, an area, a shape — has an XYZ / GPS switch on its Transform section once the environment has a geo reference. GPS swaps the position fields for Lat, Lon and Alt, so you can type a coordinate off a site plan or a survey instead of working out how many metres east of the origin it is. It is the same position either way: the scene still stores metres from the origin, and the switch only changes how you read and type it. Turning it on and off does not move anything. The fields open on GPS while the map is showing and on XYZ otherwise, which is usually what you want — once you pick one, it stays picked. Altitude is absolute, in the datum your geo reference is set to, not height above the floor. The line under the fields says which. A value is applied when you leave the field or press Enter, not as you type — every part of a coordinate is a valid coordinate somewhere else, and 4 on the way to 48.1496 is several kilometres away.

​

Position a twin from GPS

Select the twin, then in the twin editor turn on Position from GPS. It is off by default, and it requires two things:

1. The environment has a geo reference.
2. The twin declares a gps sensor — a GNSS receiver it actually has.

Options: Once on, the twin’s own position telemetry is ignored in Live mode — GPS is the single source of truth for where it is. Changing the twin’s options takes effect straight away too. Those two requirements hold for as long as GPS positioning is on, so while it is on you cannot swap the twin to an asset without a gps sensor, remove the sensor from its schema, or sync it with an asset that has none. Turn Position from GPS off first. The alternative would be a twin that quietly stops appearing in Live mode some time later, with nothing to connect it to the edit that caused it.

​

When the twin does not appear

A GPS-positioned twin with no usable position is hidden, not drawn at its saved spot, because that spot is not georeferenced and showing it would be misleading. An alert card explains which case it is. Some of these degrade the twin rather than hiding it — the card says which: The twin editor also shows a live readout — latitude, longitude, fix type, satellites and the resulting scene position — which is the quickest way to tell whether a twin is being placed by GPS or by something else.

​

Accuracy

Placement is as accurate as the fix. Consumer GNSS is typically 1.5–5 m; RTK is centimetres. The coordinate conversion itself contributes a couple of centimetres at a few hundred metres from the reference, so it is never the limiting factor. Heading is read as degrees from true north. If a device reports magnetic north without correcting for declination, its heading will be consistently off by the local declination angle.