TMA - initial scenario

[IanMayo]

This is a really easy start. Blue turns more frequently than Red, which makes it easier for Blue. We're restricting course and speed available to Red, again making solution easier. We just see if TMA can produce range and bearing.

Blue - seeker

• Travel at 8 knots, altering between 45 degs & 315 degs every 10-15 minutes
• Blue sensor produces new bearing every 5 seconds, sensor not range limited.

Red - target

• Travel at 8 knots, altering between 45 degs & 315 degs every 20-30 minutes
• Red starts 20km North of Blue.

If platforms travel more than 50km from each other, start again, with new Run-ID

Data

At each step, output:

• Run-ID
• time
• blue location, course, speed
• bearing to target
• red location, course, speed
• calculated range/bearing from blue to red

(bold figures used for training)

[IanMayo]

I've produced some bulk interaction data (attached below).

Each file contains lines like this:

100112 121314 FisherOne VC 00 10 00.00 N 000 04 29.92 E 90.00  0.00  -0.00 
100112 121314 SearchOne FA 00 00 00.00 S 000 00 22.82 E 300.00  14.00  -0.00 
;SENSOR: 100112 121314 SearchOne @A NULL 22.4 0000 "Plain Cookie" FisherOne held on Plain Cookie
100112 121319 FisherOne VC 00 10 00.00 N 000 04 30.26 E 90.00  6.00  -0.00 
100112 121319 SearchOne FA 00 00 00.67 N 000 00 21.86 E 310.32  9.00  -0.00 
;SENSOR: 100112 121319 SearchOne @A NULL 22.5 0000 "Plain Cookie" FisherOne held on Plain Cookie

This above data shows the the record of two time steps:

• FisherOne is the "truth" track. In the above sample it is travelling at 90 degs with zero speed in the first step and 6 kts in the second.
• SearchOne is the platform that's conducting the TMA. In the above sample it is travelling on 300 degs at 14 kts in the first step and 9 kts in the second.
• the ;SENSOR: is the bearing observation. In this line, it believes the target is on a bearing of 22.4 in the first step, and 22.5 on the second.

The platform positions are shown as degs, mins, secs from a point at 0,0 on the equator. While the range to the target can be calculated from the above, I do have code to put the measured range into the ;SENSOR: line above.

But, for this phase, if the TMA produces representative range and bearing that will still demonstrate success.

Here are 5 sets of runs, typically 15 hours long, with 15 second data. Successive sets introduce more variability (such as target speed)

ML_Results_1.zip ML_Results2.zip ML_Results3.zip ML_Results4.zip ML_Results5.zip