An Aergility ATLIS 1 experimental unmanned aerial vehicle (UAV), operated by Aergility Corporation and registered as N273KG, sustained substantial damage during a hard landing at Marion County Airport, Dunnellon, Florida. The flight was operated by Aergility as a flight test within airport property and was conducted under the provisions of Title 14 Code of Federal Regulations Part 91 and a Federal Aviation Administration Certificate of Authorization (COA). There were no injuries. The ATLIS is a vertical takeoff and landing (VTOL) cargo UAV currently under development by the Aergility corporation. The Aergility ATLIS UAV is equipped with six rotors with electric motors powered by batteries for vertical take-off and landing, and a turboprop propulsion engine and propeller for forward flight (see figure 1). Rotor speed (RPM) is always controlled by the autopilot to manage attitude and altitude control. The turboprop engine provides the power for forward flight and an engine driven starter/generator provides power for recharging the batteries. The accident occurred during the 16th test flight of the test program, and the fourth flight of the day. At this point in the flight test program, only hover and very slow forward flight testing had been performed with the UAV. The purpose of this flight was to examine the performance of hover when incorporating a change in center of gravity (c.g.) location within the c.g. range of the UAV. The turboprop engine for forward flight was not in use for this test, and the propeller was strapped in the “X” position. According to the operator, the UAV was being piloted in manual mode, where pilot inputs command roll angle, pitch angle, yaw rate, and vertical rate. Approximately 45 seconds into the flight, while hovering at about 15 feet above ground level (AGL), the UAV executed a planned 90-degree clockwise heading change over the takeoff location. Shortly afterward, an audible reduction in rotor RPM was noted, and the UAV began to descend. The pilot immediately input control to increase vertical rate, but the descent continued. The UAV remained level and maintained heading throughout the descent to the ground. Despite the pilot’s input to counter the descent, the descent rate was not arrested. Autopilot sensor data showed the aircraft contacted the ground at 6.8 meters/second descent rate. Figure 1. Atlis Aergility UAV. (Source: Aergility Corporation) Upon contact with the ground, the landing gear skids collapsed, and the wings deflected downward about 10-15 degrees from the normal dihedral, breaking the spars just outside the fuselage spar box. After the wing spars had broken, the wingtip rotors lifted each wing, rotating vertically around the spars until they collided directly above the fuselage, causing significant damage to the rotors, wings and booms. The fuselage experienced only minor damage. The turboprop engine was undamaged, and one of the 4 blades of the turboprop propeller was damaged. Examination of telemetry data logs and systems information revealed that all motor controllers were operating in the normal range, when the front right and left rear motor controllers simultaneously switched to the “OFF” state. The autopilot had not yet incorporated an algorithm to detect and react to a motor controller shutdown and increase thrust, and the autopilot was prioritizing roll, pitch and yaw control. As a result, the UAV was unable to maintain hover with 2 of the 6 motors shut down, yet did impact the ground in a stable pitch, roll and yaw attitude. Examination of onboard data logs and telemetry data logs of the systems information showed the data recording was active through the engine 2 and 6 shutdown, the subsequent ground impact and the resulting wing structural failure. However, the data did not reveal any error message or anomalous data related to the simultaneous shutdown of the 2 motor controllers. Examination of the hardware and electrical circuits of the accident aircraft did not reveal any evidence of failures or electrical shorts in any of the electrical systems, wiring, connections, or busses. Several conditions were identified that could have caused the motor electronic speed controllers to switch to “OFF”. The two motor controllers correctly transmitted telemetry messages following the switch to the “OFF” state indicating they were still functioning, and subsequent testing of the two motor controllers found no fault or failure in either of the two motor controllers. A short intermittent open condition could cause both controllers to switch to “OFF”, but there was no evidence in either the flight logs or hardware to conclusively identify the cause of the open condition in this accident, indicating the condition that switched off the controllers was a very brief transient electrical event. As a result of this accident, the operator/manufacturer implemented improvements to the motor controller logic to help prevent a similar event by increasing sampling of any error conditions before issuing a command to switch the motor off. For non-transient events, the controller will now reduce power before depowering when appropriate, and to re-evaluate command thresholds values. Additionally, for future flights telemetry data will include comparison with error thresholds, and multiple reporting messages when thresholds are exceeded.