The field of autonomous localization and mapping is rapidly advancing, with a focus on developing more accurate and efficient methods for estimating the position and orientation of autonomous vehicles. Recent research has explored the use of bird's-eye view (BEV) representations, which offer a metric-scaled planar workspace and facilitate the simplification of 6-DoF ego-motion to a more robust 3-DoF model. Additionally, there is a growing interest in self-supervised learning methods, which eliminate the need for ground-truth poses and offer greater scalability. Another area of research is the development of more robust and accurate trajectory prediction methods, which can handle out-of-sight objects and noisy sensor data. Noteworthy papers in this area include S-BEVLoc, which proposes a novel self-supervised framework for LiDAR global localization, and MGTraj, which introduces a multi-granularity goal-guided model for human trajectory prediction. Other notable papers include Fine-Grained Cross-View Localization, BEVTraj, and DiffVL, which propose innovative methods for cross-view localization, trajectory prediction, and visual localization, respectively.
Advances in Autonomous Localization and Mapping
Sources
S-BEVLoc: BEV-based Self-supervised Framework for Large-scale LiDAR Global Localization
MGTraj: Multi-Granularity Goal-Guided Human Trajectory Prediction with Recursive Refinement Network
Fine-Grained Cross-View Localization via Local Feature Matching and Monocular Depth Priors
BEVTraj: Map-Free End-to-End Trajectory Prediction in Bird's-Eye View with Deformable Attention and Sparse Goal Proposals
Self-supervised Learning Of Visual Pose Estimation Without Pose Labels By Classifying LED States
DisorientLiDAR: Physical Attacks on LiDAR-based Localization
Maps for Autonomous Driving: Full-process Survey and Frontiers
MFAF: An EVA02-Based Multi-scale Frequency Attention Fusion Method for Cross-View Geo-Localization
Recurrent Cross-View Object Geo-Localization
FlowDrive: Energy Flow Field for End-to-End Autonomous Driving
DiffVL: Diffusion-Based Visual Localization on 2D Maps via BEV-Conditioned GPS Denoising
BEV-ODOM2: Enhanced BEV-based Monocular Visual Odometry with PV-BEV Fusion and Dense Flow Supervision for Ground Robots
Semantic-LiDAR-Inertial-Wheel Odometry Fusion for Robust Localization in Large-Scale Dynamic Environments
Out-of-Sight Trajectories: Tracking, Fusion, and Prediction