Target Geolocation from a Small Unmanned Aircraft System

Draper Laboratory and AeroVironment, Inc. of Monrovia, CA are implementing a system to demonstrate target geolocation from a Raven-B unmanned aircraft system (UAS) as part of the U.S. Army Natick Soldier Research, Development & Engineering Center's Small UAS (SUAS) Advanced Concept Technology Demonstration (ACTD). The system is based on feature tracking, line-of-sight calculation, and Kalman filtering from Draper's autonomous vision-aided navigation code base. The system reads imagery and telemetry transmitted by the UAS and includes a user interface for specifying targets. Tests on a snapshot of on-going work indicate horizontal targeting accuracy of approximately 10 m, compared with 20-60 m for the current Raven-B targeting software operating on the same flight video/telemetry streams. This accuracy likely will be improved through further mitigation of identified error sources. This paper presents our targeting architecture, the results of tests on simulator and flight data, an analysis of remaining error, and suggestions for reducing that error.