融合动静特征的自动驾驶场景矢量化表征方法OA

Existing research on autonomous driving scene representation tends to encode sensor data into high-dimensional features to retain maximum raw information.However,this approach introduces substantial back-ground noise irrelevant to decision-making and incurs a heavy computational burden when fusing historical features to extract dynamic information.To address these challenges,this paper proposes DSVec(Dynamic-Static Vectoriza-tion),a novel vectorization scene representation method based on dynamic-static feature fusion.First,complex traf-fic scenes are abstracted into a series of structured dynamic and static vector elements.On this basis,a dynamic-static feature fusion network is designed.A variational auto-encoder(VAE)is employed to extract low-dimensional static features from semantic bird's-eye-view(BEV)representations,while historical trajectory features of dynamic obstacles are extracted via a graph-based structure.A temporal convolutional network is then used to achieve pre-cise alignment of these heterogeneous features across both spatial and temporal dimensions.Subsequently,a Trans-former decoder augmented with a category-level mask mechanism is introduced to eliminate redundant information via an attention mechanism,enabling deep fusion of static and dynamic features as well as independent decoupling and reconstruction of various vectorized elements.Finally,a soft actor-critic(SAC)deep reinforcement learning decision-making model is constructed based on the compact vectorized state space generated by DSVec,and end-to-end closed-loop validation is conducted on the CARLA high-fidelity simulation platform.Experimental results demon-strate that the proposed method achieves high accuracy in reconstructing both dynamic and static elements.Com-pared with traditional rasterization-based representation methods,DSVec significantly enhances the environmental adaptability,decision-making safety,and computational efficiency of autonomous vehicles in complex dynamic sce-narios such as roundabouts and unprotected intersections.