28 Oct 2008
- Cory's chaining experiment
There was a single flight of NexSTAR-1 today to test Cory's chaining controller. In general, the flight was a success in that nothing went wrong. However, the gains Cory chose for his control were to small, resulting in the fact that the controller was not given enough time to converge. All of the available plots can be found in the plots subfolder.
NexSTAR-1 Flight 1
Flight Time: 32.8 mins
Autonomous Flight Time: 28.1 mins
Two MNR's were deployed on Table Mountain, and NexSTAR-1 was tasked to chain between them. The idea of chaining is to have the UA find a location (the control point of the orbital controller) between the two MNR's, where the SNR of each node is maximized and equalized. At this location, the UA should be in a location where it can act as an optimal relay between the two nodes.
In inspection of the data, it turns out that the center point update gain was too small to have convergence in reasonable time. From the plots, it can be seen that if the experiment was continued for about another 10 minutes, then the two SNRs from each of the MNRs would have been balanced. It appears that at roughly every 5 min there is a increase in signal strength of the nodes in the network. This can be seen at mission times of 52 min and 56.5 min. The cause of this is not exactly known. Interestingly, on 14 Oct 2008, just the opposite occurred. There was a signal dropout every 5 mins.
Chaining of Two, Static, Ground-Based MNRs
In the first plot, you can see we the gps track of NexSTAR-1 during the ~12 min experiment. At the start of the experiment, NexSTAR-1 was up North, orbiting by MNR-64. The second plot shows the raw sampled SNR values for the MNRs and the GCS-Laptop. The solid lines in the second plot are the averages.
As the experiment ran, the controller drove the control point (center point of the orbit) towards a more central location of the two MNRs. The third plot shows that in general, the average distance between the UA and the ground based MNRs was converging. The final plot shows that on average, the controller was reducing the difference in the SNRs of the two nodes, which is the ultimate goal. Notice that according to the SNR difference plot, in about another 10 minutes, the two SNR's would have converged. In the next flight, the control point gains will be increased.