대형 언어 모델을 사용한 엣지 디바이스에서의 효율적인 주행 행동 서술 및 추론
대형 언어 모델(LLM)을 엣지 디바이스에 적용하여 주행 행동을 실시간으로 서술하고 추론할 수 있는 효율적인 프레임워크를 제안한다.
Graph Recurrent Attentive Neural Process Model for Efficient Vehicle Trajectory Prediction and Uncertainty Quantification
GRANP, a novel model combining Graph Attention Networks, LSTM, and Recurrent Attentive Neural Processes, can efficiently capture spatial-temporal relationships and quantify prediction uncertainties for vehicle trajectory forecasting.
Driving Everywhere with Large Language Model Policy Adaptation: A Powerful Tool for Seamless Traffic Rule Navigation
LLaDA, a simple yet powerful tool, enables human drivers and autonomous vehicles to adapt their driving behavior to traffic rules in new locations by leveraging the zero-shot generalizability of large language models.
Autonomous Driving Styles: Exploring User Preferences for Lateral Behavior on Rural Roads
Passengers prefer a more passive lateral driving style for autonomous vehicles on rural roads, especially under adverse weather conditions and in the presence of oncoming traffic.
Efficient Lane Change Classification and Prediction Using Action Recognition Networks
Action recognition models can efficiently extract spatial and temporal clues from video data to accurately classify and predict lane change events of surrounding vehicles in autonomous driving scenarios.
Sparse Query-Centric Paradigm for Efficient End-to-End Autonomous Driving
The authors propose a novel Sparse Query-Centric paradigm for end-to-end Autonomous Driving (SparseAD), where the sparse queries completely represent the whole driving scenario across space, time and tasks without any dense BEV representation, enabling efficient extension to more modalities and tasks.
AGENTSCODRIVER: A Large Language Model-Powered Framework for Collaborative and Lifelong Learning Autonomous Driving
AGENTSCODRIVER is a novel framework that leverages large language models to enable multiple vehicles to conduct collaborative driving with the capabilities of lifelong learning, reasoning, communication, and reflection.
UniPAD: A Universal Pre-training Paradigm for Effective 3D Representation Learning in Autonomous Driving
UniPAD is a novel self-supervised learning paradigm that leverages 3D differentiable rendering to effectively learn continuous 3D representations, enabling seamless integration into both 2D and 3D frameworks for autonomous driving tasks.
Enhancing End-to-End Autonomous Driving through Multi-Modal Prompting and Language Model Integration
Combining basic driving imitation learning with Large Language Models (LLMs) based on multi-modality prompt tokens to enhance end-to-end autonomous driving performance.
Enhancing Autonomous Driving Safety with Partial Observation Prediction (POP) Framework
A novel trajectory prediction framework called Partial Observations Prediction (POP) that employs self-supervised learning and feature distillation techniques to provide stable and accurate predictions even with limited observations.
© 2024 by Linnk AI