Real-time estimation of road slope based on multiple models and multiple data fusion

Abstract Although driving conditions of vehicles are complex, a single model is used in conventional slope estimation methods. This reduces the ability to track changes in actual road slopes in real time, resulting in low estimation accuracy. Furthermore, slope estimation based on a single estimation method are affected by braking, shifting, and sensor failure. This reduces the robustness and reliability of the slope estimation. To address these problems, we propose a slope estimation algorithm based on multi-model and multi-data fusion. First, for each estimation method (kinematics-based and dynamics-based), a two-layer interacting multiple model (TLIMM) slope estimation algorithm is developed. Second, the decision-level estimation fusion is based on the slope data estimated by the two methods. The experimental results show that the proposed TLIMM algorithm and multi-data estimation fusion method based on TLIMM can effectively improve the accuracy, robustness and reliability of the slope estimation.