Generative Lane Topology Reasoning via Autoregressive Model with Geometry Prior

2026-06-30

AI Summary

This work addresses geometric inconsistencies and structural incompleteness in lane topology inference from in-vehicle sensor observations, which arise from decoupled detection and association stages. To this end, the paper proposes TopoGPT—the first approach that integrates autoregressive generative modeling with geometric priors for lane graph generation. By tokenizing lane graphs into discrete sequences and employing a scene context encoder alongside an autoregressive Transformer, the model enables end-to-end training to produce structurally complete and geometrically consistent lane topologies. Evaluated on the OpenLane-V2 benchmark, the method achieves notable improvements of 6.4 and 11.6 points on lane-level and point-level metrics, respectively, substantially mitigating challenges posed by occlusion and connectivity inconsistencies.

Abstract

Lane topology reasoning aims to construct a lane graph from onboard sensor observations. Existing methods follow a detection and association paradigm that treats each lane instance independently, leading to geometric inconsistency at connected endpoints and incomplete graphs due to visual occlusions. To address these issues, we propose TopoGPT, a generative framework that learns the geometry prior from typical lane graph structures through autoregressive sequence modeling. Specifically, we construct a large-scale map dataset comprising 3.3M scenes. For each lane graph, a lane tokenizer serializes it into discrete tokens, while a scene context encoder converts it into a rasterized image and extracts global features as scene tokens. We pre-train an autoregressive lane sequence transformer via scene-conditioned next-token prediction, endowing the model with the geometry prior over lane graph structures. Building upon this prior, a perception adapter aligns BEV features from multi-view images with the pre-trained scene condition, transferring the learned geometry prior to sensor-based lane graph prediction. On the OpenLane-V2 benchmark, TopoGPT outperforms existing methods by an average of +6.4 on lane-level and +11.6 on point-level metrics, and produces geometrically consistent and structurally complete lane graphs.

Problem

• lane topology reasoning
• geometric inconsistency
• visual occlusions
• lane graph construction

Innovation

• autoregressive modeling
• geometry prior
• lane topology reasoning
• generative framework
• BEV perception