MapPrior: Bird's-Eye View Map Layout Estimation with Generative Models
This work addresses the challenge of uncertainty and realism in BEV map layout estimation for autonomous driving, representing an incremental improvement over existing methods.
The paper tackles the problem of generating realistic and coherent semantic map layouts in bird's-eye view perception by introducing MapPrior, a framework that combines discriminative and generative models, resulting in improved accuracy, realism, and uncertainty awareness, with significant gains in MMD and ECE scores on the nuScenes benchmark.
Despite tremendous advancements in bird's-eye view (BEV) perception, existing models fall short in generating realistic and coherent semantic map layouts, and they fail to account for uncertainties arising from partial sensor information (such as occlusion or limited coverage). In this work, we introduce MapPrior, a novel BEV perception framework that combines a traditional discriminative BEV perception model with a learned generative model for semantic map layouts. Our MapPrior delivers predictions with better accuracy, realism, and uncertainty awareness. We evaluate our model on the large-scale nuScenes benchmark. At the time of submission, MapPrior outperforms the strongest competing method, with significantly improved MMD and ECE scores in camera- and LiDAR-based BEV perception.