Read automatically from the scorecard below — every claim can be checked against it.
16 % of the score comes from initial soc error (T10, 3.74 % × weight 0.100) — improving there moves the leaderboard position most.
Accuracy holds on the never-released cell (2.17 % vs 2.28 % on the open cells) — the model generalises rather than memorising the released data.
Cold is the weak spot: 4.25 % at −20 °C against 1.81 % at 25 °C. Below zero the cell's resistance rises and the voltage curve distorts; models without temperature awareness lose the most here.
Recovers from a wrong initial SOC (3.74 % on test 10) — the estimator pulls itself back using voltage rather than trusting its starting point.
Tolerates a biased current sensor (2.50 % on test 11) — the drift a bias causes is being corrected away.
Scorecard
Average RMSE (% SOC) per test case, weighted into the leaderboard score. Lower is better.
Estimation accuracyT1All cells(reference only — every other test is a subset of it)
2.25 %
0.0000
0.0000
T2Blinded cell (m448)
2.17 %
0.1000
0.2171
T3
Key cases
The runs that separate estimators — cold and hot cycles, the blinded cell, wrong initial SOC, a biased current sensor. Each plot says why it is there.
Selected drive cycles
The thirteen cycles the original MATLAB tool plotted, one per test: blinded vs non-blinded cell (tests 1–3), drive vs charge (test 4), the three payloads on one cycle (tests 5–6), standard vs non-standard highway profiles (tests 7–8) and the temperature extremes (test 9).
m80 UDDS at -20 °C
Whole cycle · RMSE 3.47 %, max 8.4 %
Actual SOC
LSTM-Romulo
Error axis
Why this case: Test 9 — coldest condition: high resistance and a strongly non-linear OCV response — the hardest cycle for most estimators.
Wrong initial SOC (test 10)
Three cycles restarted with the estimator told the battery is at 90, 60 or 30 % while it is really full. Coulomb counting can never recover; voltage-informed estimators converge at a rate that is very visible here.
Wrong initial SOC 90 % — m80 LA92 at 25 °C
Whole cycle · RMSE 1.32 %, max 11.1 %
Actual SOC
LSTM-Romulo
Error axis
Why this case: The cycle is joined at the point where the true SOC has already fallen to 90 %, so a model that assumes a full battery at start-up begins 10 % off. What matters is how fast the estimate converges onto the truth.
Current-sensor offset (test 11)
Three cycles re-run with ±0.3 A added to the measured current. Integration-based estimators drift steadily; the plot shows whether and how fast the model corrects the bias.
Current-sensor offset -0.3 A — m1000 US06 at -10 °C
Whole cycle · RMSE 3.01 %, max 7.3 %
Actual SOC
LSTM-Romulo
Error axis
Why this case: A constant -0.3 A is added to the measured current before it reaches the model. Pure current integration drifts linearly; estimators that also use voltage should correct it.
RMSE vs. sensor offset
Test 11 — the same three cycles re-run with a constant current offset; 0 A is the unmodified run. Only ±0.3 A enters the score.
US06 at −10 °C
HWFET at 10 °C
LA92 at 40 °C
Test-case chartsThe scorecard as bar charts — tests 1–8, and RMSE against temperature (test 9).
Error by test case
Average RMS error (% SOC) for each blinded test case. Lower is better.
Error vs. ambient temperature
Test 9 — m80 cell, all blinded cycles at each chamber temperature.
All 144 cyclesEvery blinded drive cycle: per-cycle errors, and any of the plotted cycles in the time domain.
Cell
Temp.
Cycle
Duration
RMSE
MAE
MAXE
m80
-10 °C
UDDS
6.32 h
1.95
1.66
5.72
m80
-10 °C
HWFET
3.02 h
3.50
3.10
7.11
m80
-10 °C
LA92
4.74 h
1.33
1.10
4.18
m80
-10 °C
US06
2.30 h
1.17
0.92
4.82
m80
-10 °C
HWCUST
1.69 h
1.07
0.83
5.07
m80
-10 °C
HWGRADE
1.24 h
3.61
2.76
10.13
m80
-20 °C
UDDS
4.62 h
3.44
3.13
8.35
m80
-20 °C
HWFET
2.34 h
6.11
5.48
10.02
m80
-20 °C
LA92
3.54 h
4.19
3.91
9.88
m80
-20 °C
US06
1.86 h
2.88
2.65
6.44
m80
-20 °C
HWCUST
1.37 h
4.57
4.17
8.40
m80
-20 °C
HWGRADE
1.25 h
4.32
3.62
10.99
m80
0 °C
UDDS
8.75 h
1.78
1.57
4.68
m80
0 °C
HWFET
3.84 h
1.42
1.13
4.97
m80
0 °C
LA92
6.39 h
1.51
1.31
6.75
m80
0 °C
US06
2.93 h
2.51
2.07
7.96
m80
0 °C
HWCUST
2.08 h
1.62
1.36
4.54
m80
0 °C
HWGRADE
1.64 h
4.90
3.74
17.83
m80
10 °C
UDDS
11.32 h
0.97
0.82
2.26
m80
10 °C
HWFET
4.49 h
0.93
0.79
2.23
m80
10 °C
LA92
7.88 h
0.49
0.37
1.81
m80
10 °C
US06
3.42 h
0.73
0.60
2.47
m80
10 °C
HWCUST
2.42 h
0.86
0.71
1.80
m80
10 °C
HWGRADE
1.80 h
1.51
1.24
5.06
m80
25 °C
UDDS
16.13 h
5.21
3.52
11.99
m80
25 °C
HWFET
5.52 h
0.84
0.65
3.15
m80
25 °C
LA92
10.58 h
0.73
0.60
2.29
m80
25 °C
US06
4.22 h
0.89
0.69
3.73
m80
25 °C
HWCUST
2.86 h
1.54
1.33
2.97
m80
25 °C
HWGRADE
2.33 h
1.67
1.40
4.60
m80
40 °C
UDDS
12.40 h
1.54
1.22
4.87
m80
40 °C
HWFET
5.41 h
1.46
1.12
4.50
m80
40 °C
LA92
9.07 h
1.56
1.22
5.14
m80
40 °C
US06
4.17 h
1.45
1.15
4.50
m80
40 °C
HWCUST
2.95 h
1.16
0.95
3.23
m80
40 °C
HWGRADE
2.73 h
1.76
1.28
8.46
m448
-10 °C
UDDS
6.23 h
1.98
1.74
5.23
m448
-10 °C
HWFET
3.01 h
3.02
2.77
5.97
m448
-10 °C
LA92
4.59 h
1.23
1.00
4.14
m448
-10 °C
US06
2.25 h
1.36
1.07
4.67
m448
-10 °C
HWCUST
1.69 h
0.61
0.49
4.24
m448
-10 °C
HWGRADE
1.22 h
4.08
3.02
12.30
m448
-20 °C
UDDS
4.56 h
2.99
2.69
8.39
m448
-20 °C
HWFET
2.34 h
5.32
4.75
9.08
m448
-20 °C
LA92
3.46 h
3.76
3.56
8.30
m448
-20 °C
US06
1.80 h
2.56
1.92
6.24
m448
-20 °C
HWCUST
1.39 h
3.93
3.54
7.57
m448
-20 °C
HWGRADE
1.23 h
4.30
3.68
9.34
m448
0 °C
UDDS
8.60 h
1.73
1.52
5.39
m448
0 °C
HWFET
3.83 h
1.03
0.86
3.40
m448
0 °C
LA92
6.16 h
1.84
1.58
7.44
m448
0 °C
US06
2.87 h
3.70
3.02
10.22
m448
0 °C
HWCUST
2.09 h
1.74
1.43
4.79
m448
0 °C
HWGRADE
1.51 h
5.79
4.27
21.71
m448
10 °C
UDDS
11.08 h
1.15
0.94
2.88
m448
10 °C
HWFET
4.47 h
0.47
0.38
2.00
m448
10 °C
LA92
7.60 h
0.64
0.49
2.62
m448
10 °C
US06
3.35 h
0.92
0.72
3.45
m448
10 °C
HWCUST
2.42 h
0.47
0.37
1.50
m448
10 °C
HWGRADE
1.68 h
2.17
1.78
7.23
m448
25 °C
UDDS
15.59 h
5.80
3.96
13.21
m448
25 °C
HWFET
5.46 h
0.85
0.70
2.43
m448
25 °C
LA92
10.06 h
0.72
0.58
2.71
m448
25 °C
US06
4.07 h
0.91
0.71
4.53
m448
25 °C
HWCUST
2.86 h
1.81
1.47
4.78
m448
25 °C
HWGRADE
2.19 h
1.80
1.44
5.31
m448
40 °C
UDDS
12.09 h
1.47
1.15
4.87
m448
40 °C
HWFET
5.21 h
1.50
1.12
4.77
m448
40 °C
LA92
8.79 h
1.54
1.17
5.68
m448
40 °C
US06
4.06 h
1.50
1.18
4.88
m448
40 °C
HWCUST
2.92 h
1.23
1.00
3.49
m448
40 °C
HWGRADE
2.32 h
2.24
1.66
8.99
m448-N
-10 °C
UDDS
10.80 h
1.16
0.98
4.58
m448-N
-10 °C
HWFET
3.85 h
2.83
2.60
6.43
m448-N
-10 °C
LA92
6.66 h
0.92
0.68
5.08
m448-N
-10 °C
US06
2.70 h
1.51
1.22
4.73
m448-N
-10 °C
HWCUST
1.89 h
3.10
2.94
5.37
m448-N
-10 °C
HWGRADE
1.36 h
4.12
3.17
11.21
m448-N
-20 °C
UDDS
8.67 h
3.93
3.40
8.60
m448-N
-20 °C
HWFET
3.09 h
5.72
5.13
10.31
m448-N
-20 °C
LA92
5.21 h
2.11
1.75
8.87
m448-N
-20 °C
US06
2.23 h
2.48
2.24
6.60
m448-N
-20 °C
HWCUST
1.65 h
6.56
5.86
11.60
m448-N
-20 °C
HWGRADE
1.50 h
6.30
5.37
14.62
m448-N
0 °C
UDDS
12.79 h
3.24
3.01
7.66
m448-N
0 °C
HWFET
4.47 h
0.88
0.68
3.41
m448-N
0 °C
LA92
7.99 h
2.52
2.24
9.32
m448-N
0 °C
US06
3.22 h
3.35
2.57
11.36
m448-N
0 °C
HWCUST
2.31 h
2.36
2.06
5.32
m448-N
0 °C
HWGRADE
1.65 h
5.86
4.28
22.36
m448-N
10 °C
UDDS
13.92 h
1.91
1.70
4.05
m448-N
10 °C
HWFET
4.84 h
0.38
0.29
1.63
m448-N
10 °C
LA92
8.80 h
1.03
0.84
3.50
m448-N
10 °C
US06
3.56 h
0.99
0.78
3.65
m448-N
10 °C
HWCUST
2.53 h
0.47
0.38
1.30
m448-N
10 °C
HWGRADE
1.79 h
2.21
1.81
9.00
m448-N
25 °C
UDDS
15.61 h
0.91
0.67
2.98
m448-N
25 °C
HWFET
5.47 h
0.71
0.57
2.66
m448-N
25 °C
LA92
10.09 h
0.74
0.59
2.34
m448-N
25 °C
US06
4.08 h
0.84
0.66
3.65
m448-N
25 °C
HWCUST
2.86 h
1.57
1.36
3.72
m448-N
25 °C
HWGRADE
2.30 h
1.83
1.44
5.82
m448-N
40 °C
UDDS
16.75 h
1.54
1.23
4.88
m448-N
40 °C
HWFET
5.89 h
1.53
1.18
4.59
m448-N
40 °C
LA92
11.03 h
1.62
1.27
5.50
m448-N
40 °C
US06
4.44 h
1.51
1.20
4.90
m448-N
40 °C
HWCUST
3.13 h
1.27
1.05
3.59
m448-N
40 °C
HWGRADE
2.48 h
2.22
1.66
11.52
m1000
-10 °C
UDDS
5.91 h
1.92
1.72
5.15
m1000
-10 °C
HWFET
2.91 h
3.64
3.42
7.31
m1000
-10 °C
LA92
4.23 h
1.43
1.21
5.19
m1000
-10 °C
US06
2.09 h
1.75
1.38
5.60
m1000
-10 °C
HWCUST
1.67 h
2.47
2.29
4.84
m1000
-10 °C
HWGRADE
1.20 h
4.73
3.41
16.57
m1000
-20 °C
UDDS
4.31 h
3.89
3.60
9.15
m1000
-20 °C
HWFET
2.27 h
5.87
5.26
9.77
m1000
-20 °C
LA92
3.20 h
3.86
3.54
9.56
m1000
-20 °C
US06
1.74 h
4.14
3.48
9.43
m1000
-20 °C
HWCUST
1.37 h
4.52
4.01
8.71
m1000
-20 °C
HWGRADE
1.22 h
6.03
5.25
13.55
m1000
0 °C
UDDS
8.09 h
1.58
1.33
5.33
m1000
0 °C
HWFET
3.70 h
1.60
1.40
4.45
m1000
0 °C
LA92
5.60 h
1.79
1.53
6.94
m1000
0 °C
US06
2.66 h
4.28
3.57
12.10
m1000
0 °C
HWCUST
2.05 h
2.29
2.01
7.92
m1000
0 °C
HWGRADE
1.36 h
7.34
5.58
31.93
m1000
10 °C
UDDS
10.51 h
0.89
0.72
2.19
m1000
10 °C
HWFET
4.37 h
0.86
0.75
2.17
m1000
10 °C
LA92
7.06 h
0.54
0.44
2.90
m1000
10 °C
US06
3.11 h
0.98
0.77
4.09
m1000
10 °C
HWCUST
2.41 h
0.72
0.60
1.79
m1000
10 °C
HWGRADE
1.52 h
2.77
2.18
12.18
m1000
25 °C
UDDS
14.69 h
1.05
0.88
2.77
m1000
25 °C
HWFET
5.37 h
0.93
0.76
2.55
m1000
25 °C
LA92
9.30 h
0.84
0.69
2.81
m1000
25 °C
US06
3.79 h
1.11
0.90
4.72
m1000
25 °C
HWCUST
2.83 h
1.87
1.65
4.44
m1000
25 °C
HWGRADE
1.90 h
2.08
1.61
7.94
m1000
40 °C
UDDS
11.69 h
1.37
1.08
4.53
m1000
40 °C
HWFET
5.15 h
1.34
0.99
4.29
m1000
40 °C
LA92
8.33 h
1.42
1.08
5.42
m1000
40 °C
US06
3.89 h
1.35
1.06
4.84
m1000
40 °C
HWCUST
2.91 h
1.00
0.81
2.81
m1000
40 °C
HWGRADE
2.17 h
2.55
1.87
9.53
m80 UDDS at -20 °C
Whole cycle · RMSE 3.47 %, max 8.4 %
Actual SOC
LSTM-Romulo
Error axis
One hour of padded data precedes every cycle in the evaluator and is excluded from the error metrics. Charts are down-sampled for display (peaks preserved); the full 1 Hz data is in the traces download below.
Score historyEvery evaluation attempt and every change to how this submission is scored; the current score is the last row.
How a lstm worksThe canonical structure for this model family; the author's description below gives the specific architecture.
LSTM — standardized view
Data-driven, recurrent
Governing equation
it,ft,ot=σ(Wxt+Uht−1+b)
Downloads
PDF report (summary, every test case with the arithmetic, key plots, per-cycle table) · results as JSON · full 1 Hz traces of all 195 runs as a MATLAB v7 file (0.01 % SOC steps; readable with scipy.io.loadmat, see the readme variable inside).
LSTM-Romulo.zip (36.6 KB) — deleted after evaluation
Visibility
Public
Evaluator
socbench-eval-0.1.0/matlab
Completed
Aug 31, 2026, 08:50 p.m. EDT
Submission ID
cmthv7uv90001l204qrcdwf8p
Non-blinded cells
2.28 %
0.1000
0.2280
Operating conditionsT4Charging
1.87 %
0.1000
0.1873
T580 kg payload
2.14 %
0.0333
0.0715
T5448 kg payload, HVAC on
2.17 %
0.0667
0.1447
T6448 kg payload, HVAC off
2.29 %
0.0667
0.1523
T51000 kg payload
2.41 %
0.0333
0.0803
T7Standard drive cycles
1.97 %
0.1000
0.1973
T8Non-standard drive cycles
2.81 %
0.1000
0.2813
T9−20 °C
4.25 %
0.0167
0.0709
T9−10 °C
2.11 %
0.0167
0.0351
T90 °C
2.29 %
0.0167
0.0382
T910 °C
0.91 %
0.0167
0.0152
T925 °C
1.81 %
0.0167
0.0302
T940 °C
1.49 %
0.0167
0.0248
RobustnessT10Initial SOC error
3.74 %
0.1000
0.3739
T11Current sensor offset
2.50 %
0.1000
0.2502
Weighted error = Σ (weight × RMSE)
2.40 %
Max error 31.9 % — largest instantaneous |estimate − truth| over every blinded cycle (on m1000 HWGRADE at 0 °C — plotted under Key cases for new evaluations); not scoredComplexity 10 ±1 · Extreme+ (informational, not scored)
Ahnaf Rahman · McMaster University