Skip to content

MicroAir MTF-01P

The MTF-01P is the sensor that makes indoor position hold possible on this drone. It combines two sensors in one small module:

  • an optical flow camera that tracks how the ground texture moves under the drone (horizontal velocity), and
  • a TOF LiDAR rangefinder that measures the distance straight down (height above ground).

Indoors there is no GPS, so this pair replaces it: flow supplies the EKF's horizontal velocity, the LiDAR supplies the height reference the flow scaling needs. How exactly the data feeds the EKF is covered in the sensors section; this page is about the module as a piece of hardware.

The MicoAir MTF-01P is sold as a compact 2-in-1 optical-flow range sensor for drones and autonomous aircraft; the module officially supports ArduPilot, PX4, INAV and FMT.

The module has a single UART. It can speak two protocols:

Protocol Used by Notes
MSP (MultiWii Serial Protocol) Betaflight/INAV Factory default
MAVLink ArduPilot What we need — sends OPTICAL_FLOW and DISTANCE_SENSOR messages

The module ships in MSP mode

Out of the box the MTF-01P talks MSP and ArduPilot will see nothing on the port. It must be switched to MAVLink at 115200 baud once, using the MicroAir assistant software over a CP2102 USB-UART adapter (included in the team toolkit). The full procedure — wiring of the adapter, assistant settings, and every FC parameter — is in MTF-01P Configuration.

On the flight controller side the module is wired to SERIAL5 (MAVLink1, 115200 baud) and registered as FLOW_TYPE = 5 (MAVLink) and RNGFND1_TYPE = 10 (MAVLink) — see the configuration page for the complete parameter set on ArduCopter 4.6.3.

Mounting requirements

Both sensors look through the bottom of the module, so mounting is not optional detail — it decides whether the sensor works at all:

  • Facing straight down. The rangefinder orientation is configured as downward (RNGFND1_ORIENT = 25); a tilted mount skews both the height reading and the flow scaling.
  • Completely free optical path. Nothing may protrude into the field of view — no zip ties, wires, prop-guard edges or landing-gear parts. The flow camera also needs visible floor texture and light to produce anything useful.
  • Low over the ground when landed. On our airframe the lens sits roughly 2 cm above the floor with the drone at rest.

A reading of 0.02 m on the ground is CORRECT

Do not "fix" a 2 cm reading while the drone stands on the floor — that is the true lens-to-floor distance, and it matters: the rangefinder minimum must be configured below it (RNGFND1_MIN_CM = 1 on 4.6.3, not the 20 cm default). With a higher minimum the driver reports "out of range low" on the ground, the EKF gets no terrain height, optical flow cannot be scaled and arming fails with "Need Position Estimate". In our crash analysis the sensor itself performed flawlessly — status good on 3441 of 3441 samples, tracking the climb from 0.02 m to 4.94 m (incident report).

Status: final mount still to be designed

The mechanical mount that fixes the MTF-01P under the frame (together with the AI-camera and servo mounts) is still open as of 2026-08-22 — the sensor has so far been configured and bench-tested, and validated in the 2026-08-21 flight logs.

Known limitations

  • No position estimate while grounded. At 2 cm height the flow camera sees almost no usable texture, so a position estimate only becomes available once airborne — a chicken-and-egg problem for autonomous takeoff that we solve with a pilot handover (--takeover) in the companion code.
  • Floor dependent. Featureless, dark or glossy floors degrade flow quality; our logs show flow quality between 45 and 113 on the hall floor.
  • Range limits. Configured usable rangefinder window is 1 cm to 8 m (RNGFND1_MIN_CM/RNGFND1_MAX_CM).
  • Our EKF height source, and mitigated. We run the LiDAR as the EKF's altitude source (EK3_SRC1_POSZ = 2). The task asks us to use the LiDAR and the optical flow for altitude hold and position hold. This is the configuration that crashed us on 2026-08-21 when flown without mitigations, so we fly it only under the safety protocol (ground-drift preflight, rangefinder-gated takeover, in-flight EKF-vs-rangefinder cross-check, RNGFND1_GNDCLEAR = 5, the parameter's own minimum in Mission Planner rather than the true ~2 cm mount height). Moving the EKF height source back to the barometer (EK3_SRC1_POSZ = 1) is an option we have deliberately not taken: the rangefinder is the sensor the task is about, and the 2026-08-25 SITL work showed the on-ground non-fusion was the RNGFND1_MIN_CM validity floor rather than the source choice itself. It stays on the table if the real aircraft disagrees, and under either setting the LiDAR keeps the jobs the assignment asks for: it scales the optical flow, it is the low-altitude terrain reference, and it is the independent height witness behind the takeover gate and the in-flight cross-check. The barometer stays as an independent witness in the logs. Details in the incident report and the sensors section.