Incorporating explainable AI and user-friendly interfaces is crucial for building trust and situation awareness in autonomous vehicles, as it enables end-users to understand the reasoning behind the vehicle's decisions and actions.
SLAM (Simultaneous Localization and Mapping) technology plays a crucial role in enabling accurate environment perception, vehicle positioning, and intelligent decision-making for automated lane change functionality in autonomous driving systems.
CurbNet introduces a novel framework for curb detection using point cloud segmentation, achieving exceptional results and setting new benchmarks in autonomous driving technology.
Efficient radar-camera fusion for accurate 3D object detection.
ICP-Flow, a learning-free flow estimator using the Iterative Closest Point (ICP) algorithm, outperforms state-of-the-art baselines in scene flow estimation for autonomous driving.
LiDAR-based 3D object detection benefits from semantic features obtained through 3D semantic segmentation, improving performance for cars.
Lightning NeRF introduces an efficient hybrid scene representation for autonomous driving, significantly improving novel view synthesis performance and reducing computational overheads.
CalibFormer proposes an end-to-end network for automatic LiDAR-camera calibration, achieving high accuracy and robustness.
Proposing ACC-DA for collaborative perception in autonomous driving to minimize transmission delay, enhance data reconstruction, and align data distribution.
OccFusion integrates cameras, lidar, and radar to enhance 3D occupancy prediction accuracy and robustness.