No fatalities

20 Feb 2024: Apellix B1 Washing Drone — Drone Cleaning Group — Orlando, FL

Orlando, FL, United States

On 20 Feb 2024, an Apellix B1 Washing Drone operated by Drone Cleaning Group was involved in an aviation accident near Orlando, FL. No fatalities were reported. Investigators recorded the probable cause as: A failure of the sUAS’s magnetometer while operating in close proximity to commercial air conditioning equipment, which resulted in the sUAS’s erratic maneuvering. This summary draws on records from NTSB.

SourcesNTSBPrimary reportUpdated 1778583330Data APIEditorial standards

An Apellix B1 sUAS (FA33N7WLTR) operating under Part 107 lost control near Orlando, Florida on February 20, 2024, causing serious injuries to the remote pilot when he manually disarmed the battery. The flight controller data revealed erratic responses and magnetometer faults.

Accident Summary

On February 20, 2024, at 1438 eastern standard time, an Apellix B1 small unmanned aircraft system (sUAS), registration FA33N7WLTR, was involved in an accident near Orlando, Florida. The remote pilot-in-command (RPIC) sustained serious injuries. The flight was conducted under Title 14 Code of Federal Regulations Part 107.

Flight Purpose and Sequence

The RPIC stated that the flight aimed to power wash a hospital roof; the accident occurred on the second day of cleaning. During a repositioning flight, after takeoff at about 3 feet above the takeoff point, the sUAS began to lose directional control. The RPIC attempted to control it using the remote control, but the sUAS responded erratically to inputs. The RPIC tried to disarm the motors via the control station, but the sUAS did not respond. Determining a loss of control, the RPIC grabbed the sUAS and manually unplugged the battery, sustaining several lacerations on his arms, hands, and legs, and partially severing a finger.

Location and Equipment

Review of onboard GPS data and aerial imagery showed the accident flight occurred on top of the hospital roof, near a large rectangular object consistent with a commercial air conditioning unit.

Training and Emergency Procedures

The RPIC reported receiving condensed training from the manufacturer due to prior experience with other power washing sUASs. During training, the RPIC was told that bringing both control sticks down and inward would shut down the motors in an emergency. The manufacturer's operations manual (Section 2.4, “Fly-Away (Complete Loss of Control)”) stated that the RPIC should first attempt to send a return-to-home command, then activate the flight termination system (if equipped). If these failed, the RPIC was to note the direction of travel, maintain visual line of sight, and warn others. However, the sUAS operations and wash system user manuals contained no instructions on how to disarm the sUAS in an emergency or how to engage the flight termination system. According to the manufacturer, the accident sUAS was configured to shut down by the RPIC bringing both sticks inward at a 45-degree angle (the same command used to start the motors), which would cut all power to motors and flight processor.

Post-Accident Actions and Data

Before notifying the NTSB, the operator returned the sUAS to the manufacturer for repair. The manufacturer downloaded the flight controller's internal datalogger, repaired the sUAS by replacing propellers, two arms with motors, radio, receiver, and lower landing gear parts, and returned it to the operator. The sUAS was not provided to an FAA inspector before repairs. The manufacturer provided the datalogger data to the NTSB.

Data analysis revealed the sUAS was airborne for about 19 seconds. It locked onto 14 GPS satellites and took off in assisted takeoff mode at 1437:58, then switched to GPS Positioning (P-GPS) mode. In the first 5 seconds, the remote controller (RC) commanded full elevator input but pitch remained relatively unchanged; full aileron input resulted in a +9.9° roll. Then the elevator was commanded to 0% then back to +100%, and pitch changed to -10.1°. Rudder command was -50%, and heading changed from 124.7° to 84.9°. No fault alerts appeared in the first 5 seconds.

From 1438:03 to 1438:07, elevator inputs varied, pitch changed to -9.7° and later varied. Aileron inputs did not fully correspond with roll: at 1438:04, aileron at +100% gave +13.1° roll; when aileron dropped to +32%, roll increased to +47.1°. Rudder at 0% while heading turned left from 58.4° to 16.9°, then right to 79.9°, then left to 54.7°. A fault “magn_heading_err_large” was logged, and the sUAS switched to “Sport” mode. Another fault “COMPASS(2): fault on, over_large” appeared, and GPS satellite count dropped from 12 to 5.

From 1438:08 to 1438:10, elevator was mostly at 0% with a brief +54%, pitch varied between -9.2° and +20.4°. Aileron varied between +23% and +100%, roll between 0.5° and 29.3°. Rudder at 0% while heading turned right from 55.4° to 183.1°. Additional faults included “COMPASS(2): fault on, interfere.” GPS satellite locks increased from 5 to 11.

In the final 6 seconds, elevator input was -100% but pitch remained at about +2°. Aileron at 2% while roll was +65°. Rudder at +65% while heading stayed around 164°. Faults “[CTRL]: fault on, height_ctrl_fail” and “[L-RC]craft ctrl failed!!!” appeared. The recording ended with elevator at -100%, pitch +1.6°, aileron +2%, roll +64.1°, rudder +65%, heading 164.7°.

Examination of control stick inputs showed that the sticks were never moved to the disarm position during the flight. The data indicated a continuous RC connection with no lost link alerts. The manufacturer stated that none of the recorded magnetometer errors were provided as feedback to the RPIC.

Contributing factors

MalfunctionPilotManufacturer