153b95a88f
Keep Auto Learn evidence current
2026-11-01 13:13:40 -06:00
5235f58e07
Keep FLM trial input tapes current between motion deliveries
2026-10-31 13:39:45 -05:00
6d2e382e92
Renew FLM from already received motion during FIFO ingest
2026-10-31 13:06:52 -05:00
70f523edc5
Renew owned FLM trials when admitted queued input becomes ready
2026-10-31 10:46:02 -05:00
c1f37c8314
Renew owned FLM trials when admitted queued input becomes ready
2026-10-31 10:26:39 -05:00
15ed9799cb
Renew owned FLM trials when admitted queued input becomes ready
2026-10-31 09:43:22 -05:00
b8200e7bab
Revisit pending FLM motion before another evidence drain
2026-10-31 08:19:02 -05:00
0d8c81e294
Renew an owned FLM trial before draining queued evidence
2026-10-30 21:49:47 -05:00
efc8527820
Restore measured FLM receiver after queue scheduling regressions
2026-10-30 21:27:10 -05:00
d421abeee1
Yield FLM queues only when actual motion needs service
2026-10-30 21:05:45 -05:00
7da9cea0f6
Yield FLM evidence queues before fresh pose and trial delivery
2026-10-30 20:41:23 -05:00
f4dde57839
Keep FLM input reception responsive and numerical work off Controls
2026-10-30 15:09:04 -05:00
0d795930ff
Keep the FLM live receiver at normal CPU priority
2026-10-29 16:58:34 -05:00
8fcd77c953
Recover FLM learning windows and deliver current trial evidence
2026-10-29 00:44:04 -05:00
e095ab6b71
Retain real FLM source packets for bounded frequency accounting
2026-10-28 14:01:27 -05:00
0fcc393380
Pace optional FLM receiver on motion without relaxing evidence guards
2026-10-28 12:33:01 -05:00
9f4fb7dbb4
Preserve FLM evidence during overload and remove unused wakeups
2026-10-28 09:11:23 -05:00
487dfc4bf4
FLM: preserve fresh evidence across paired queue drains
...
Merge bounded control/output FIFO heads using original publisher headers. Preserve original age, native, source, overflow and malformed-input checks; retain producer order after joins.
Validated with actual pinned cloud dependencies and fresh-head regression tests.
2026-10-27 17:01:24 -05:00
6103e0395b
Preserve FLM learning windows across unchanged steering repeats
...
The d5 Ioniq 6 logs repeatedly reset source tracking when Card repeats a command that already arrived on time. Discard only a recent unchanged repeat after the existing native CAN verifier succeeds; add no sample, application time or permission.
Use independently admitted pending CarState brackets to complete original poses without waiting for that command to reach Card. Classify stale control source clocks as transient stale evidence, preserving the original bounded study deadline. Replace Python character loops with equivalent strict text predicates.
Validation: focused regression suite and paced d5 replay with original Card command references; actual controller/CAN trial application and worsening-trial rollback. No timing/acceptance threshold changes or claimed road improvement.
2026-10-26 21:55:26 -05:00
ec25e2330a
Consume pending native evidence before FLM collection
2026-10-26 11:41:40 -05:00
0f00621c98
Coalesce refreshed FLM receipts with scheduled tracking
2026-10-26 09:34:09 -05:00
0e5e5574c4
Publish refreshed FLM collection receipts on tracking
2026-10-26 09:09:55 -05:00
f3bbd38e81
Wake Auto Learn on fresh motion and native evidence
2026-10-25 21:05:53 -05:00
3c740d2050
Allow conflated car state snapshots in Auto Learn
2026-10-25 13:55:50 -05:00
4c250a6a06
Restore Auto Learn source admission after GPU startup fix
2026-10-24 21:55:54 -05:00
24126eafdc
Keep source learning progressing after interrupted trials
2026-10-24 09:45:44 -05:00
d2f3acc248
Keep calibrated configuration across learner receipt gaps
2026-10-23 20:10:59 -05:00
0022fde1cb
Fix Auto Learn collection recovery after evidence gaps
2026-10-23 17:36:10 -05:00
a5eda8a63a
Fix Auto Learn authorization and tracking delivery timing
2026-10-23 17:13:21 -05:00
cdb7e7a5d4
Reject missing Auto Learn withdrawal receipts
2026-10-22 21:39:39 -05:00
95155234e8
Publish explicit Auto Learn source withdrawals
2026-10-22 21:37:28 -05:00
15a78b9b33
Maintain Auto Learn source delivery cadence
2026-10-22 21:29:34 -05:00
ca2373a9a0
Refresh Auto Learn inputs after queued decoding
2026-10-22 20:14:03 -05:00
62dc081f70
Keep Auto Learn source publication responsive
2026-10-22 19:10:12 -05:00
1f66903899
Recover Auto Learn after calibration delivery gaps
2026-10-22 11:11:11 -05:00
b151ca11cb
Global FLM Auto Learn
2026-10-21 23:28:39 -05:00
3a8dbe75f1
Repair CI Contracts
2026-10-19 00:47:35 -05:00
6ce281bf5f
Fix FLM evidence and proposal lifecycle
2026-10-18 18:30:31 -05:00
07adfe819e
Complete FLM learning pipeline
2026-10-17 23:26:55 -05:00
5b12042b88
Fix lateral learning intake
2026-10-17 20:02:38 -05:00
2cd7728be2
FLM Auto-Weave
2026-10-17 18:38:27 -05:00
198a9fca63
FLM Training Results
2026-10-17 10:53:16 -05:00
53309097e9
Fix Auto Learn collection renewal
2026-10-16 17:21:48 -05:00
abb9646499
FLM Calibration Lifecycle
...
Keep bounded automatic learning collection through valid live calibration refinements. Retire generations on resets or invalid calibration, preserve live physical transforms and existing controller tunes, and validate fitted coefficient bounds before publication. Exercise the current controller and historical wire paths.
2026-10-16 09:47:39 -05:00
7e63c553a4
FLM Auto-Weave
2026-10-16 08:46:39 -05:00
200632aabd
StarPilot Auto Learn
...
Learn bounded feed-forward corrections on eligible torque controllers behind Use StarPilot Auto Learn, preserving installed tunes and native safety limits.
Co-authored-by: Danny <dannyirwin31@yahoo.com >
Co-authored-by: Dom <168790843+firestar5683@users.noreply.github.com >
Co-authored-by: Harald Schäfer <harald.the.engineer@gmail.com >
Co-authored-by: Prabhaav Pillai <prabhaav@msu.edu >
Co-authored-by: dirwin31 <83434411+dirwin31@users.noreply.github.com >
Co-authored-by: firestarsdog <229254897+firestarsdog@users.noreply.github.com >
Co-authored-by: Intelli <contact@intelli.software >
Co-authored-by: haraschax <6804392+haraschax@users.noreply.github.com >
Co-authored-by: pharmacomaniac <blittle65@gmail.com >
Co-authored-by: whoisdomi <dcantu83@gmail.com >
2026-10-14 20:02:39 -05:00
firestar5683