cs.CVSep 12, 2025

BEVTraj: Map-Free Multimodal Trajectory Prediction with Sensor-Derived BEV Context and Sparse Goal Reasoning

Authors: Minsang Kong, Myeongjun Kim, Sang Gu Kang, Hejiu Lu, Yupeng Zhong, Sang Hun Lee

Organizations: Department of Automobile and IT Convergence, Kookmin University, 77 Jeongneung-ro, Seongbuk-gu, Seoul 02707, Republic of Korea · Department of Automotive Engineering, Kookmin University, 77 Jeongneung-ro, Seongbuk-gu, Seoul 02707, Republic of Korea · Graduate School of Automobile and Mobility, Kookmin University, 77 Jeongneung-ro, Seongbuk-gu, Seoul 02707, Republic of Korea

Abstract

Trajectory prediction is a key task in autonomous driving, where a model must forecast plausible future motions of surrounding agents from dynamic histories and scene context. Many recent methods use high-definition (HD) maps to provide explicit road topology and semantic constraints, but such maps are costly to maintain, geographically limited, and difficult to update in changing environments. This paper presents Bird's-Eye View Trajectory Prediction (BEVTraj), a map-free trajectory prediction framework that directly uses BEV features constructed from raw multi-sensor data. Since sensor-derived BEV features are dense and only implicitly structured, BEVTraj employs deformable attention to selectively aggregate agent-relevant spatial context from sparse locations rather than processing the entire BEV grid uniformly. In addition, a Sparse Goal Candidate Proposal module combines overlapping mode-wise anchor clusters with attention-based goal prediction to generate compact, scene-adaptive goal hypotheses. These goals are decoded and iteratively refined into multimodal trajectories, while trajectory-aware mode scoring and temporal consistency modeling improve confidence estimation and suppress abrupt, physically implausible refinement updates. Experiments on nuScenes and Argoverse 2 Sensor show that BEVTraj achieves competitive results under a map-free setting without using pre-built HD maps, while ablation studies and qualitative analyses illustrate the roles of BEV context aggregation and sparse goal reasoning. The source code is available at https://github.com/Kongminsang/bevtraj.

Explore similar work

CardsList
  1. SIPTraj: Map-Free End-to-End Trajectory Prediction via Physics-Guided Scene Interaction

    Aug 1, 2026Feifei Liu, Zejun Wei, Haozhe Wang +6Multi-Agent Trajectory PredictionAutonomous Driving

  2. ObjectGuidedMap: An Object-Aware Model for Robust Bird's Eye View Map Prediction in High-Occlusion Scenarios

    Oct 30, 2026Christian Bohn, Harsh Yadav, Klaus Friedrichs +2Autonomous Driving PerceptionMulti-Task Learning

  3. MC-DeTra: Motion-Consistent Joint Object Detection and Socially-Aware Trajectory Forecasting in Bird's-Eye-View Images

    Sep 10, 2026Vladislav Diuzhev, Dmitry YudinMulti-Agent Trajectory PredictionAutonomous Driving