Subject: C1001 (SEN0623) — Is posture data available in fall detection mode?
refael 2026-08-02 21:31:29 101 Views1 Replies Hi,
I'm using the C1001 in fall detection mode (eFallingMode) with the DFRobot_HumanDetection library on an ESP32.
The product description states that the sensor "recognizes human postures through point cloud imaging algorithms" and can detect standing, sitting and lying. However, in fall detection mode the library only exposes:
smHumanData(eHumanPresence)smHumanData(eHumanMovement) — None/Still/ActivesmHumanData(eHumanMovingRange)getFallData(eFallState)getFallData(estaticResidencyState)There is no posture output (standing / sitting / lying).
My questions:
Does the sensor internally compute posture in fall detection mode, and is there a UART command or register that exposes it — even if the library doesn't wrap it?If not, is the posture recognition mentioned in the product description only available in sleep mode (eSleepMode)?My use case: detecting a person who slides down slowly (legs give way, gradual descent) rather than falling fast. The fall flag doesn't trigger for slow descents. I want to distinguish "lying on the floor" from "sitting still in an armchair" — without posture data both look identical to my logic.Is there any recommended approach for this with the C1001?
Thanks!
refael
The library header doesn't list any standing/sitting/lying posture enum for fall mode — the human data query only covers presence, motion, body movement, trajectory switch, seated horizontal distance and motion horizontal distance. So the posture recognition in the product description is likely internal to the fall algorithm and not surfaced as a separate output. Sleep mode has the same limitation — the human data there only covers presence, movement, moving range and distance. No posture labels either. For your slow-collapse use case, the sensor is tuned for rapid speed/distance/posture changes, so a gradual descent may not trigger the fall flag. You could try dialing down the fall sensitivity and break height config to catch slower drops, but no guarantee. To tell "lying on floor" from "sitting still", the track function gives you x/y coordinates and the seated horizontal distance query might help infer position. Combined with the height cumulative time value you can build a rough height-based heuristic, but it's not the same as a posture label.
Jason.Miao 
