Week 105 - Resumen de métricas
Ver convenciones y definiciones en Week 102.
Test 0 - Caso canónico
Town02
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 98.7% (222/225) | 0.000 | 7.090 | 55.33 | 0.233 |
| $D_{0.15}^{B}$ | 99.1% (223/225) | 0.099 | 6.493 | 61.09 | 0.397 |
| $D_{0.30}^{B}$ | 96.9% (218/225) | 1.000 | 8.798 | 55.64 | 0.348 |
| $D_{0.50}^{B}$ | 99.1% (223/225) | 0.000 | 8.076 | 57.16 | 0.514 |
| $D_{0.75}^{B}$ | 49.1% (110/224) | 1.618 | 24.255 | 46.73 | 0.580 |
| $D_{0.15}^{C}$ | 97.3% (219/225) | 0.151 | 7.598 | 54.14 | 0.324 |
| $D_{0.30}^{C}$ | 97.3% (219/225) | 0.000 | 6.667 | 59.42 | 0.329 |
| $D_{0.50}^{C}$ | 99.6% (224/225) | 0.009 | 7.094 | 56.28 | 0.276 |
| $D_{0.75}^{C}$ | 60.9% (137/225) | 0.533 | 11.759 | 54.53 | 0.390 |
| $D^{M}_{30/30/40}$ | 65.8% (148/225) | 0.000 | 6.946 | 64.86 | 0.257 |
Town11
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 100.0% (30/30) | 0.000 | 8.233 | 71.72 | 0.361 |
| $D_{0.15}^{B}$ | 100.0% (30/30) | 0.000 | 12.300 | 66.70 | 0.356 |
| $D_{0.30}^{B}$ | 100.0% (30/30) | 0.000 | 21.367 | 72.12 | 0.429 |
| $D_{0.50}^{B}$ | 96.7% (29/30) | 0.000 | 24.621 | 62.33 | 0.409 |
| $D_{0.75}^{B}$ | 100.0% (29/29) | 0.000 | 33.276 | 65.98 | 0.523 |
| $D_{0.15}^{C}$ | 100.0% (30/30) | 0.000 | 11.933 | 63.37 | 0.341 |
| $D_{0.30}^{C}$ | 96.7% (29/30) | 0.000 | 12.345 | 67.40 | 0.312 |
| $D_{0.50}^{C}$ | 100.0% (30/30) | 0.000 | 5.767 | 63.99 | 0.231 |
| $D_{0.75}^{C}$ | 100.0% (30/30) | 0.000 | 15.400 | 54.12 | 0.277 |
| $D^{M}_{30/30/40}$ | 100.0% (30/30) | 0.000 | 9.667 | 56.35 | 0.311 |
Csonthomisi
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 0.0% (0/24) | — | — | — | — |
| $D_{0.15}^{B}$ | 31.8% (7/22) | 0.000 | 31.714 | 45.62 | 0.361 |
| $D_{0.30}^{B}$ | 84.0% (21/25) | 0.000 | 42.619 | 29.28 | 0.424 |
| $D_{0.50}^{B}$ | 84.0% (21/25) | 0.000 | 32.571 | 40.37 | 0.400 |
| $D_{0.75}^{B}$ | 80.0% (20/25) | 0.000 | 57.000 | 34.55 | 0.459 |
| $D_{0.15}^{C}$ | 29.2% (7/24) | 0.000 | 41.143 | 48.09 | 0.509 |
| $D_{0.30}^{C}$ | 20.0% (5/25) | 0.000 | 40.200 | 43.99 | 0.459 |
| $D_{0.50}^{C}$ | 84.0% (21/25) | 0.000 | 32.571 | 43.12 | 0.419 |
| $D_{0.75}^{C}$ | 95.8% (23/24) | 0.000 | 29.391 | 39.73 | 0.364 |
| $D^{M}_{30/30/40}$ | 96.0% (24/25) | 0.000 | 38.833 | 39.32 | 0.436 |
Las métricas se calculan únicamente sobre los recorridos exitosos. El símbolo «—» indica que el modelo no completó ningún recorrido.
Test 1 – Diferentes posiciones de inicio
Town02
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 100.0% (225/225) | 0.000 | 8.707 | 54.71 | 0.262 |
| $D_{0.15}^{B}$ | 98.2% (221/225) | 0.104 | 8.136 | 60.75 | 0.417 |
| $D_{0.30}^{B}$ | 94.7% (213/225) | 1.005 | 10.404 | 55.27 | 0.381 |
| $D_{0.50}^{B}$ | 97.3% (219/225) | 0.000 | 10.050 | 56.65 | 0.526 |
| $D_{0.75}^{B}$ | 40.4% (91/225) | 1.044 | 18.055 | 47.81 | 0.600 |
| $D_{1.00}^{B}$ | 0.0% (0/225) | — | — | — | — |
| $D_{0.15}^{C}$ | 96.9% (218/225) | 0.147 | 9.156 | 53.90 | 0.346 |
| $D_{0.30}^{C}$ | 96.4% (217/225) | 0.000 | 8.295 | 59.39 | 0.337 |
| $D_{0.50}^{C}$ | 74.2% (167/225) | 0.006 | 8.449 | 56.51 | 0.307 |
| $D_{0.75}^{C}$ | 45.3% (102/225) | 0.578 | 8.824 | 53.51 | 0.392 |
| $D^{M}_{30/30/40}$ | 81.8% (184/225) | 0.000 | 8.457 | 57.87 | 0.294 |
Las métricas se calculan únicamente sobre los recorridos exitosos. El símbolo «—» indica que el modelo no completó ningún recorrido.
Town11
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 93.3% (28/30) | 0.000 | 4.929 | 76.78 | 0.289 |
| $D_{0.15}^{B}$ | 100.0% (30/30) | 0.000 | 49.733 | 77.53 | 0.327 |
| $D_{0.30}^{B}$ | 100.0% (30/30) | 0.000 | 27.067 | 61.02 | 0.427 |
| $D_{0.50}^{B}$ | 100.0% (30/30) | 0.000 | 26.067 | 61.10 | 0.415 |
| $D_{0.75}^{B}$ | 100.0% (30/30) | 0.000 | 34.300 | 62.06 | 0.546 |
| $D_{0.15}^{C}$ | 100.0% (30/30) | 0.000 | 11.833 | 75.76 | 0.296 |
| $D_{0.30}^{C}$ | 100.0% (30/30) | 0.000 | 13.567 | 66.89 | 0.321 |
| $D_{0.50}^{C}$ | 100.0% (30/30) | 0.000 | 7.172 | 65.95 | 0.228 |
| $D_{0.75}^{C}$ | 93.1% (27/29) | 0.000 | 15.481 | 55.66 | 0.299 |
| $D^{M}_{30/30/40}$ | 86.7% (26/30) | 0.000 | 11.308 | 57.62 | 0.345 |
Csonthomisi
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 7.7% (2/26) | 0.000 | 36.000 | 52.31 | 0.515 |
| $D_{0.15}^{B}$ | 26.9% (7/26) | 0.000 | 32.857 | 45.56 | 0.347 |
| $D_{0.30}^{B}$ | 81.5% (22/27) | 0.000 | 42.955 | 28.32 | 0.437 |
| $D_{0.50}^{B}$ | 77.8% (21/27) | 0.000 | 37.143 | 40.50 | 0.403 |
| $D_{0.75}^{B}$ | 91.3% (21/23) | 0.000 | 62.095 | 35.23 | 0.455 |
| $D_{0.15}^{C}$ | 20.8% (5/24) | 0.000 | 43.400 | 46.11 | 0.527 |
| $D_{0.30}^{C}$ | 33.3% (9/27) | 0.000 | 39.778 | 43.97 | 0.445 |
| $D_{0.50}^{C}$ | 88.9% (24/27) | 0.000 | 36.583 | 41.88 | 0.419 |
| $D_{0.75}^{C}$ | 96.3% (26/27) | 0.000 | 30.923 | 39.46 | 0.378 |
| $D^{M}_{30/30/40}$ | 96.3% (26/27) | 0.000 | 42.308 | 38.24 | 0.439 |
Test 2 – Condiciones iniciales aleatorias
Town02
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 80.0% (96/120) | 0.000 | 7.750 | 55.61 | 0.259 |
| $D_{0.15}^{B}$ | 90.2% (203/225) | 0.000 | 10.182 | 55.70 | 0.510 |
| $D_{0.30}^{B}$ | 98.2% (221/225) | 0.579 | 2.027 | 43.96 | 0.314 |
| $D_{0.50}^{B}$ | 99.1% (223/225) | 0.000 | 2.327 | 45.19 | 0.380 |
| $D_{0.75}^{B}$ | 32.9% (74/225) | 1.230 | 13.338 | 36.41 | 0.560 |
| $D_{0.15}^{C}$ | 98.2% (221/225) | 0.032 | 2.000 | 43.51 | 0.255 |
| $D_{0.30}^{C}$ | 72.0% (162/225) | 0.117 | 2.074 | 49.70 | 0.263 |
| $D_{0.50}^{C}$ | 36.0% (81/225) | 0.000 | 8.506 | 56.08 | 0.289 |
| $D_{0.75}^{C}$ | 34.7% (78/225) | 0.974 | 9.179 | 52.14 | 0.376 |
Town11
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 70.0% (21/30) | 0.000 | 0.238 | 67.77 | 0.225 |
| $D_{0.15}^{B}$ | 73.3% (22/30) | 0.000 | 2.682 | 60.71 | 0.355 |
| $D_{0.30}^{B}$ | 73.3% (22/30) | 0.000 | 10.455 | 62.40 | 0.495 |
| $D_{0.50}^{B}$ | 73.3% (22/30) | 0.000 | 6.318 | 56.33 | 0.286 |
| $D_{0.75}^{B}$ | 100.0% (30/30) | 0.000 | 21.267 | 60.05 | 0.447 |
| $D_{0.15}^{C}$ | 80.0% (24/30) | 0.000 | 5.625 | 59.12 | 0.345 |
| $D_{0.30}^{C}$ | 76.7% (23/30) | 0.000 | 3.391 | 62.93 | 0.281 |
| $D_{0.50}^{C}$ | 73.3% (22/30) | 0.000 | 1.773 | 59.85 | 0.192 |
| $D_{0.75}^{C}$ | 75.9% (22/29) | 0.000 | 7.000 | 54.78 | 0.261 |
| $D^{M}_{30/30/40}$ | 66.7% (20/30) | 0.000 | 5.550 | 56.30 | 0.495 |
Csonthomisi
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 33.3% (10/30) | 0.000 | 27.100 | 51.96 | 0.486 |
| $D_{0.15}^{B}$ | 39.1% (9/23) | 0.000 | 32.333 | 48.82 | 0.452 |
| $D_{0.30}^{B}$ | 10.0% (3/30) | 0.000 | 40.667 | 29.36 | 0.432 |
| $D_{0.50}^{B}$ | 42.9% (12/28) | 0.000 | 28.917 | 41.07 | 0.387 |
| $D_{0.75}^{B}$ | 30.0% (9/30) | 0.000 | 60.111 | 35.42 | 0.460 |
| $D_{0.15}^{C}$ | 38.1% (8/21) | 0.000 | 37.625 | 47.41 | 0.477 |
| $D_{0.30}^{C}$ | 20.8% (5/24) | 0.000 | 34.200 | 43.47 | 0.390 |
| $D_{0.50}^{C}$ | 53.3% (16/30) | 0.000 | 31.812 | 42.39 | 0.401 |
| $D_{0.75}^{C}$ | 32.1% (9/28) | 0.000 | 29.556 | 39.15 | 0.361 |
| $D^{M}_{30/30/40}$ | 26.7% (8/30) | 0.000 | 35.875 | 39.02 | 0.403 |
Test 3 – Diferentes velocidades iniciales
Town02
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 98.6% (218/221) | 0.000 | 7.110 | 59.71 | 0.236 |
| $D_{0.15}^{B}$ | 76.7% (168/219) | 0.310 | 13.226 | 63.77 | 0.410 |
| $D_{0.30}^{B}$ | 89.0% (195/219) | 1.021 | 17.164 | 56.89 | 0.365 |
| $D_{0.50}^{B}$ | 78.2% (172/220) | 0.058 | 15.942 | 59.75 | 0.505 |
| $D_{0.75}^{B}$ | 33.6% (74/220) | 1.541 | 23.338 | 48.62 | 0.563 |
| $D_{0.15}^{C}$ | 97.7% (215/220) | 0.186 | 7.595 | 58.04 | 0.319 |
| $D_{0.30}^{C}$ | 91.8% (202/220) | 0.139 | 6.480 | 63.11 | 0.301 |
| $D_{0.50}^{C}$ | 97.6% (161/165) | 0.006 | 6.969 | 60.01 | 0.277 |
| $D_{0.75}^{C}$ | 43.6% (24/55) | 0.500 | 7.000 | 55.06 | 0.375 |
Town11
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 64.3% (18/28) | 0.000 | 7.278 | 75.54 | 0.314 |
| $D_{0.15}^{B}$ | 85.7% (24/28) | 0.000 | 14.333 | 67.87 | 0.338 |
| $D_{0.30}^{B}$ | 96.4% (27/28) | 0.000 | 23.593 | 74.95 | 0.474 |
| $D_{0.50}^{B}$ | 100.0% (28/28) | 0.000 | 28.250 | 63.59 | 0.425 |
| $D_{0.75}^{B}$ | 96.4% (27/28) | 0.000 | 33.667 | 65.82 | 0.520 |
| $D_{0.15}^{C}$ | 77.8% (21/27) | 0.000 | 12.238 | 65.97 | 0.317 |
| $D_{0.30}^{C}$ | 57.1% (16/28) | 0.000 | 13.625 | 69.04 | 0.296 |
| $D_{0.50}^{C}$ | 82.1% (23/28) | 0.000 | 6.957 | 66.78 | 0.222 |
| $D_{0.75}^{C}$ | 59.3% (16/27) | 0.000 | 17.000 | 56.52 | 0.292 |
Csonthomisi
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 16.7% (4/24) | 0.000 | 26.000 | 48.16 | 0.500 |
| $D_{0.15}^{B}$ | 70.8% (17/24) | 0.000 | 18.412 | 43.42 | 0.306 |
| $D_{0.30}^{B}$ | 91.7% (22/24) | 0.000 | 34.091 | 27.70 | 0.360 |
| $D_{0.50}^{B}$ | 87.0% (20/23) | 0.000 | 25.500 | 39.98 | 0.378 |
| $D_{0.75}^{B}$ | 73.9% (17/23) | 0.000 | 44.294 | 30.90 | 0.478 |
| $D_{0.15}^{C}$ | 42.1% (8/19) | 0.000 | 32.375 | 44.73 | 0.498 |
| $D_{0.30}^{C}$ | 61.1% (11/18) | 0.000 | 31.727 | 41.96 | 0.442 |
| $D_{0.50}^{C}$ | 82.6% (19/23) | 0.000 | 26.211 | 44.56 | 0.432 |
| $D_{0.75}^{C}$ | 95.8% (23/24) | 0.000 | 28.826 | 41.15 | 0.378 |
| $D^{M}_{30/30/40}$ | 88.5% (23/26) | 0.000 | 31.696 | 39.80 | 0.441 |
Test 4 – Perturbación estándar
Town02
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 88.0% (198/225) | 0.025 | 8.020 | 54.34 | 0.232 |
| $D_{0.15}^{B}$ | 30.2% (68/225) | 0.221 | 7.279 | 56.81 | 0.420 |
| $D_{0.30}^{B}$ | 71.1% (160/225) | 0.994 | 10.463 | 53.92 | 0.350 |
| $D_{0.50}^{B}$ | 62.2% (140/225) | 0.129 | 9.764 | 54.53 | 0.531 |
| $D_{0.75}^{B}$ | 20.0% (45/225) | 0.978 | 18.356 | 47.63 | 0.577 |
| $D_{0.15}^{C}$ | 82.2% (185/225) | 0.076 | 8.778 | 52.91 | 0.337 |
| $D_{0.30}^{C}$ | 79.6% (179/225) | 0.179 | 7.788 | 56.15 | 0.342 |
| $D_{0.50}^{C}$ | 91.1% (205/225) | 0.054 | 8.415 | 54.60 | 0.290 |
| $D_{0.75}^{C}$ | 47.1% (106/225) | 0.509 | 10.717 | 52.19 | 0.398 |
| $D^{M}_{30/30/40}$ | 61.6% (138/224) | 0.065 | 7.848 | 56.79 | 0.279 |
| $D^{M}_{70/20/10}$ | 58.2% (131/225) | 0.076 | 9.000 | 52.06 | 0.416 |
Town11
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 0.0% (0/30) | — | — | — | — |
| $D_{0.15}^{B}$ | 83.3% (25/30) | 0.000 | 29.080 | 46.91 | 0.365 |
| $D_{0.30}^{B}$ | 90.0% (27/30) | 0.000 | 28.778 | 57.46 | 0.434 |
| $D_{0.50}^{B}$ | 96.7% (29/30) | 0.000 | 31.069 | 54.32 | 0.426 |
| $D_{0.75}^{B}$ | 73.3% (22/30) | 0.000 | 47.500 | 57.24 | 0.506 |
| $D_{0.15}^{C}$ | 0.0% (0/30) | — | — | — | — |
| $D_{0.30}^{C}$ | 70.0% (21/30) | 0.000 | 22.143 | 61.20 | 0.328 |
| $D_{0.50}^{C}$ | 100.0% (30/30) | 0.000 | 19.167 | 57.46 | 0.228 |
| $D_{0.75}^{C}$ | 73.3% (22/30) | 0.000 | 25.409 | 50.94 | 0.285 |
| $D^{M}_{30/30/40}$ | 100.0% (29/29) | 0.000 | 13.379 | 53.61 | 0.468 |
Las métricas se calculan únicamente sobre los recorridos exitosos. El símbolo «—» indica que el modelo no completó ningún recorrido.
Csonthomisi
| Modelo | Successful Runs | Collisions | Lane Invasions | Speed (km/h) | Dev Mean (m) |
|---|---|---|---|---|---|
| $D^{A}$ | 23.3% (7/30) | 0.000 | 27.857 | 54.34 | 0.515 |
| $D_{0.15}^{B}$ | 65.5% (19/29) | 0.000 | 38.632 | 42.63 | 0.377 |
| $D_{0.30}^{B}$ | 76.7% (23/30) | 0.000 | 52.609 | 27.82 | 0.450 |
| $D_{0.50}^{B}$ | 96.6% (28/29) | 0.000 | 37.500 | 37.49 | 0.395 |
| $D_{0.75}^{B}$ | 3.3% (1/30) | 0.000 | 20.000 | 35.28 | 0.441 |
| $D_{0.15}^{C}$ | 33.3% (10/30) | 0.000 | 49.300 | 44.31 | 0.516 |
| $D_{0.30}^{C}$ | 31.0% (9/29) | 0.000 | 50.667 | 42.12 | 0.441 |
| $D_{0.50}^{C}$ | 79.3% (23/29) | 0.000 | 39.435 | 40.32 | 0.421 |
| $D_{0.75}^{C}$ | 0.0% (0/30) | — | — | — | — |
| $D^{M}_{30/30/40}$ | 0.0% (0/30) | — | — | — | — |
Análisis
-
Burbuja ofrece buenos resultados en Town02, pero su robustez depende de la condición inicial. Obtiene 98.7 %, 100 % y 98.6 % de éxito en los Tests 0, 1 y 3, con desviaciones medias entre 0.233 y 0.262 m. En el Test 2, con condiciones iniciales aleatorias, completa 96/120 recorridos —80.0 %—, entre 18.6 y 20 puntos porcentuales menos que en esas pruebas. La tasa de éxito disminuye, aunque la desviación media de los recorridos exitosos se mantiene en 0.259 m.
-
En el Test 2 de Town02, DAgger al 50 % y Noise al 15 % destacan con distintos compromisos de conducción. DAgger al 50 % alcanza 99.1 % de éxito —223/225—, con cero colisiones en los recorridos exitosos, 2.327 invasiones de carril y 0.380 m de desviación media. Noise al 15 % completa 221/225 —98.2 %— y presenta menos invasiones —2.000— y menor desviación —0.255 m—, aunque registra 0.032 colisiones por recorrido exitoso. Frente al 80.0 % de Burbuja, las diferencias de éxito son de 19.1 y 18.2 puntos porcentuales, respectivamente. Ambas configuraciones reducen las invasiones respecto a las 7.750 de Burbuja y circulan a menor velocidad media: 45.19 y 43.51 km/h, frente a 55.61 km/h. Las mejoras observadas coinciden, por tanto, con cambios en la velocidad de conducción. La comparación se basa en 120 recorridos para Burbuja y 225 para cada una de estas configuraciones.
-
Una proporción mayor de datos adicionales no garantiza mejores resultados. En el Test 2 de Town02, Noise alcanza 98.2 % con el 15 %, pero desciende a 72.0 %, 36.0 % y 34.7 % con el 30 %, 50 % y 75 %. DAgger mejora hasta el 50 %, pero cae al 32.9 % con el 75 %, acompañado de 1.230 colisiones y 13.338 invasiones por recorrido exitoso. En el Test 1 de este mismo mapa, DAgger al 100 % no completa ninguno de los 225 recorridos. Sin embargo, DAgger al 75 % alcanza 100 % en el Test 2 de Town11: el efecto de la proporción cambia según el escenario.
-
Town11 muestra un compromiso entre finalización y precisión lateral. En el Test 2, DAgger al 75 % completa 30/30 recorridos frente a 21/30 de Burbuja, pero registra 21.267 invasiones y 0.447 m de desviación, frente a 0.238 y 0.225 m de Burbuja. En el Test 3, DAgger al 50 % alcanza 100 % de éxito, frente al 82.1 % de Noise al 50 %, aunque presenta más invasiones —28.250 frente a 6.957— y mayor desviación —0.425 frente a 0.222 m—. Noise al 50 % obtiene la menor desviación media de las configuraciones disponibles en los cinco test de Town11, pero no lidera la finalización en todos ellos.
-
Noise al 50 % alcanza la mayor tasa de éxito bajo perturbación estándar en Town02 y Town11. En Town02 completa 205/225 recorridos —91.1 %— frente a 198/225 —88.0 %— de Burbuja. Esa mejora coincide con mayor desviación —0.290 frente a 0.232 m— y más colisiones —0.054 frente a 0.025— entre los recorridos exitosos. En Town11 completa 30/30, mientras que Burbuja y Noise al 15 % no completan ninguno. Frente a DAgger al 50 %, que completa 29/30, Noise al 50 % también reduce las invasiones de 31.069 a 19.167 y la desviación de 0.426 a 0.228 m.
-
En Csonthomisi, las configuraciones destacadas cambian entre pruebas. Noise al 75 % alcanza 95.8 %, 96.3 % y 95.8 % en los Tests 0, 1 y 3, frente a 0 %, 7.7 % y 16.7 % de Burbuja. Sin embargo, en el Test 2 obtiene solo 32.1 %; el mejor resultado disponible corresponde a Noise al 50 %, con 53.3 % —16/30—. Esta última configuración supera en 20 puntos porcentuales a Burbuja, aunque todavía falla en 14 de los 30 recorridos. Las condiciones iniciales aleatorias siguen siendo una dificultad para todas las configuraciones evaluadas en este mapa.
-
DAgger al 50 % destaca bajo perturbación estándar en Csonthomisi. Completa 28/29 recorridos —96.6 %—, frente a 23/29 —79.3 %— de Noise al 50 %. También registra menos invasiones —37.500 frente a 39.435— y menor desviación —0.395 frente a 0.421 m—, a una velocidad media inferior: 37.49 frente a 40.32 km/h. DAgger supera a Noise en finalización en las tres proporciones comparables de este test: 15 %, 30 % y 50 %. Este resultado limita cualquier conclusión general sobre una superioridad de Noise ante perturbaciones.
-
Magic Mix no presenta una ventaja consistente en las comparaciones disponibles. En el Test 0 de Csonthomisi, la mezcla 30/30/40 alcanza 96.0 % —24/25—, un porcentaje cercano al 95.8 % —23/24— de Noise al 75 %. En el Test 1 ambos completan 26/27, pero Magic Mix presenta más invasiones —42.308 frente a 30.923— y mayor desviación —0.439 frente a 0.378 m—. En el Test 4 de Town02, las mezclas 30/30/40 y 70/20/10 alcanzan 61.6 % y 58.2 %, por debajo de Burbuja y Noise al 50 %.