A collision prediction framework integrating scenario-based approach with data-driven approach is proposed to enhance the safety of autonomous driving vehicles as well as advanced driver assistance systems. No matter howthe autonomous driving is intelligent, it is inevitable to consider malfunction or faults of sensors, actuators, and processors, thus resulting in the collision. To address these issues, several studies have been proposed to improve performance based on model-based or data-driven approaches. However, there are several challenges in terms of the scarcity of accident data and the lack of explainability of deep neural networks. To overcome the limits of both approaches, an integrated framework that includes trajectory prediction, threat assessment, and decision-making based on convolutional neural network (CNN) for collision prediction is introduced. For more detail, both trajectory prediction based on Kalman filter and probabilistic threat metric are added in the form of a simplified bird’s eye view (SBEV), which is the input to the network. In the development of the proposed algorithm, pre-crash simulation data and experimental data have been employed. A comparative study shows that the proposed algorithm outperforms the model-based algorithm on simulation data containing safety-critical scenarios. Furthermore, it outperforms the data-driven algorithm on experimental data.

In this paper, the strategy decision algorithm for automatic evasive steering (AES) integrating a convolution neural network (CNN) with a physics-based threat assessment is proposed. Five collision avoidance or mitigation strategies, including evasive steering, lane change, and steering to shoulder stop are considered for the strategy decision. Although there are many model-based or data-driven approaches for collision avoidance in the literature, a new decision method integrating data-driven classification based on CNN with both threat assessment and prediction techniques is proposed to improve reliability as well as accuracy. First, a set of abstracted images in a bird eye’s view including the threat assessment and trajectory prediction information are generated. More specifically, a few collision indexes and interaction multiple model-unscented Kalman filter are used respectively for threat assessment and prediction. Once a stack of the images so called predicted semantic map corresponding to each collision avoidance strategy are generated, the decision classification based on CNN follows to choose an appropriate strategy for AES. Finally, the proposed decision algorithm is trained and validated through typical safe scenario data coming from field operation tests (FOT) and safety-critical scenario data via simulations.

At General Motors, we envision a future with zero crashes, zero emissions and zero congestion:

A collision prediction framework integrating scenario-based approach with data-driven approach is proposed to enhance the safety of autonomous driving vehicles as well as advanced driver assistance systems. No matter howthe autonomous driving is intelligent, it is inevitable to consider malfunction or faults of sensors, actuators, and processors, thus resulting in the collision. To address these issues, several studies have been proposed to improve performance based on model-based or data-driven approaches. However, there are several challenges in terms of the scarcity of accident data and the lack of explainability of deep neural networks. To overcome the limits of both approaches, an integrated framework that includes trajectory prediction, threat assessment, and decision-making based on convolutional neural network (CNN) for collision prediction is introduced. For more detail, both trajectory prediction based on Kalman filter and probabilistic threat metric are added in the form of a simplified bird’s eye view (SBEV), which is the input to the network. In the development of the proposed algorithm, pre-crash simulation data and experimental data have been employed. A comparative study shows that the proposed algorithm outperforms the model-based algorithm on simulation data containing safety-critical scenarios. Furthermore, it outperforms the data-driven algorithm on experimental data.