Evaluation of Model-based Real-time Latency Compensation for Unmanned Aircraft Remote Control

Kurzfassung

For advanced flight testing of novel Unmanned Aircraft Systems (UAS) based on a synthetic vision, special skills and properties are required for direct First Person View (FPV) remote control. Necessary wireless data links and computer based data processing will introduce latency into the system. A UAS demonstration system has previously been built at the German Aerospace Center (DLR). For this system it was observed that the overall latency of the system reaches a magnitude where Pilot Induced Oscillation (PIO) occurs. In addition, bandwidth limitations result in low sampling rates in the downlink, leading to jerking flight information displays and visualizations. Due to those influences, handling qualities are fairly poor and will further decrease for Beyond Visual Line Of Sight (BVLOS) operations. To provide a pilot with smooth, accurate, and up-to-date visual cues, a compensation approach is developed that accounts for the full combination of: Delays, low sample rates, unavailable aircraft state measurements, modelling uncertainties, disturbances and sensor noises. An evaluation of three model-based methods is performed in this work, supported by a pilot-in-the-loop simulation test campaign. The measured accuracy of the compensated output as well as the subjective handling quality surveys show very promising results.