Figures of Decision-Level Sensor Fusion on the CAN Bus for Autonomous Vehicles: A Standardized Evaluation of Seven Methods

0.0 48.5 50.0 32.5 16.0 48.0 2.5 34.0 35.0 45.0 44.5 44.5 47.5 82.0 86.0 94.0 50.0 78.0 34.0 95.0 90.5 91.5 97.5 102.0 89.0 84.5 82.5 80.0 98.0 95.0 92.0 91.5 87.0 182.5 98.0 93.5 94.5 14.5 65.5 53.0 60.0 100.5 99.5 105.5 96.5 8.5 42.5 7.5 32.5 31.0 31.5 32.0 33.5 4.5 34.0 16.0 41.5 34.5 36.0 42.0 29.5 32.0 52.0 31.0 56.5 49.5 53.5 51.0 59.0 39.0 58.5 38.0 70.0 66.5 73.0 70.5 78.5 56.0 97.5 56.5 62.5 70.0 60.5 60.0 199.0 77.5 73.0 60.5 72.0 102.0 53.0 159.0 57.5 64.5 64.0 70.0 67.0 132.5 165.0 96.5 68.5 78.0 75.0 80.5 84.5 Rule Based - Bayesian - Dempster-Shafer - XGBoost - SmallMLP - Sequence - SmHybrid - Near-Miss Rate (NMR) by Model and Scenario Scenario/Class Model Source(s): Authors’ own creation Figure 14. NMR comparison across all scenarios