All three systems had a clean canopy opening and behaved as predicted switching from one phase of the GN&C algorithm to another. Although the touchdown accuracy was not the main objective for this first set of experiments, the three deployed systems demonstrated only 55m circular error probable (CEP), which happened to be twice more accurate than any other existing autonomously guided ADS.
After these initial drops, an additional series of tests was performed in Yuma, AZ (October of 2008 and May of 2009), Camp Roberts, CA (February and August of 2009, May and August of 2010), Kingman, AZ (May of 2009) and Marina, CA (August and September of 2009), which enabled tuning Snowflake's guidance and control algorithms to achieve unprecedented and unbeatable accuracy. To this end, Fig.Ea features the results of series of drops performed in May of 2009 with a 30m CEP, and Fig.Eb – performed in May of 2010 with a 10m CEP!
Figure E. Snowflake accuracy as of May of 2009 (a) and May of 2010 (b).
The key features that allow Snowflake to achieve these excellent results are:
- Real-time generation of an inertial reference trajectory based on the current (constantly updated) estimates of ADS parameters and winds/atmosphere (Fig.F outlines different phases of such a trajectory);
- Real-time optimization of the final turn into the wind (for soft landing) with an update rate of as high as 100Hz;
- Advanced tracking algorithm based on the model predictive control;
- Communication / networking with the target weather station(s) and between multiple descending Snowflakes to share winds/atmosphere information (this version is called Snowflake-N and features below 10m CEP accuracy).
Figure F. Snowflake guidance strategy.
Due to these features, Snowflake ADS exhibits about the same trajectory every time it is dropped (that distinguishes it from all other existing aerial delivery systems), and softly lands into the wind (or any other predetermined direction if a specific tactical scenario, especially for multiple systems, calls for it). (Figure G presents a couple of actual trajectories.) Communication with the miniature target weather station(s) and between themselves, and capability to access the descending system from anywhere in the world (networking) makes Snowflake-N ADS a unique system to be used in a variety of novel applications. (Figure H shows the same trajectory as in Fig.Gb but viewable on-line from anywhere in the world at Google Earth.)