-
Notifications
You must be signed in to change notification settings - Fork 1
Radar Placement
Radar placement directly affects coordinate accuracy, stationary-presence retention, false detections, and multi-device fusion quality. Verify the physical position and orientation before compensating with software zones.
NAMO obtains spatial coordinates from the LD2450-series primary radar. The primary radar provides target position, motion, target count, room and zone state, and observations for multi-device fusion.
The optional LD2410C is an assist sensor that can maintain an already established presence session. It cannot start presence by itself or create coordinates, motion, target count, or room state. If coordinates jump or appear in the wrong place, inspect the primary radar installation before changing LD2410C sensitivity.
Choose a position that satisfies as many of these conditions as possible:
- The primary radar has a clear view of normal occupied positions and travel paths.
- The enclosure, cables, metal parts, and power equipment do not obstruct the radar face.
- The radar does not unintentionally cover a wide area through a door, glass partition, or thin wall.
- The radar and enclosure can be mounted without movement or tilt.
- USB power and UART wiring can remain stable.
- The dashboard remains accessible for orientation and status checks after installation.
NAMO uses a two-dimensional floorplan coordinate system. Mounting height and vertical tilt are not automatically corrected by the floorplan, so test detection at the real height and angle before permanent mounting. A placement tested against actual seats, walking paths, and furniture is safer than applying one universal mounting height to every room.
[photo: comparison of a clear radar face and an incorrect installation obstructed by a cable or metal part]
After fixing the radar temporarily, verify its direction on the local radar map first.
- Walk forward and backward in front of the radar.
- Move left and right relative to the radar.
- Confirm that the target moves in the same directions on the screen.
- Place the radar body at its real floorplan position and rotate its handle toward the real forward direction.
- Repeat the same path and check the floorplan coordinates.
If the direction is reversed or skewed, do not compensate by moving detection zones. Correct the radar position and rotation first so rooms, zones, and peer-device coordinates use the same reference.
mmWave radar can detect people through glass and some thin walls. An open glass door does not remove every effect from the surrounding glass, frame, applied film, and reflective objects.
- A radar outside a room may produce less stable coordinates when observing a person through glass.
- A seated person whose body is only partly visible may jump or disappear.
- People in a corridor or adjacent room may become unintended targets.
- Drawing a wall on the floorplan does not change the physical radar's penetration behavior.
To determine whether glass or a wall is responsible, compare the same position with a primary radar that has an unobstructed view. If only the obstructed sensor shows large coordinate movement, the installation environment and raw observation are more likely causes than the fusion algorithm.
First adjust the position or direction to reduce coverage of unintended spaces. Use a small insensitive or exclusion zone only for persistent false detections that cannot be solved physically.
[photo: comparison between observing a room through glass and observing it directly from inside]
Multiple devices can complement occluded views when they observe the same space from different directions. Devices placed at nearly the same position and angle are useful for stability comparison, but provide less coverage diversity.
- Verify the intended area on each radar independently first.
- Save every device's real position and orientation on the shared floorplan.
- Do not assume coordinates match until position alignment is complete.
- If one sensor covers too much adjacent space, review its placement before trying to hide the result through fusion settings.
- Test with the expected number of people when the space is normally shared.
Even when two radars are mounted close together for comparison, distinguish each local target from the fused target. Fusion tuning cannot repair a local coordinate that is already unstable at one radar.
- Temporarily secure the radar at the intended final position.
- Observe the empty space for persistent targets or repeated false detections.
- Walk normal routes and verify forward, backward, left, and right movement.
- Sit or stand still at the actual occupied positions.
- Cross doors and detection boundaries naturally.
- If multiple people must be detected, test with the expected occupancy.
- When the result is acceptable, secure the radar and save its floorplan position and rotation.
- Configure only the detection, exit, insensitive, or exclusion zones that are still needed.
Do not change placement, direction, radar sensitivity, and zones at the same time. Change one item at a time so the effect can be identified.
Check the mount, glass or furniture between the person and radar, partial body occlusion, and nearby reflective objects. If another sensor directly observes the same position, compare their local coordinates.
Check whether the radar sees an adjacent space through glass, an open door, or a thin wall. Do not assume the floorplan outline blocks raw detections. Adjust placement first, then apply a filter only where necessary.
First confirm that the primary radar has a useful view of the person. The LD2410C only assists an established presence session and does not replace missing coordinates. Tune the assist sensor only after checking physical placement and primary-radar behavior.
Check whether the split already exists on each device's local radar map. If it does, inspect reflections, view boundaries, and sensor settings. If local radars show one target but fusion shows more, verify position alignment and each device's floorplan placement.
- The radar body is rigidly mounted.
- No metal, cable, or enclosure structure obstructs the radar face.
- Movement on the local radar map matches real movement.
- Floorplan position and rotation match the physical installation.
- Detection remains usable at normal seats and travel paths.
- The empty and adjacent spaces have been checked for persistent false detections.
- For multiple devices, both local observations and fused results have been checked.
- Zones and sensitivity were adjusted only after physical placement was verified.