A novel vision-language framework specifically designed to enhance scene understanding in autonomous driving scenarios
Explore technology by industry and role
Newest additions first. Explore past and current notes by industry and role.
All notes
Newest additions first. Explore past and current notes by industry and role.
Systematically evaluate generalizability of five key LLM modules in motion generation
Introduces a lightweight latent probe that enables early failure detection with calibrated accuracy and minimal overhead.
Point-plane projections enable effective feature learning from 2D representations using only LiDAR data
Context-aware reliability learning and agreement-based fusion using game-theoretic team credit
ADPerf generates realistic point cloud data to expose increased latency in obstacle detection modules.
A single unified architecture efficiently processes diverse sensor data using linear group RNNs.
Introduces a temporal alternation policy to improve closed-loop robustness without increasing model size or data
Both users and items generate self-descriptions, enabling symmetric communication and resolving information asymmetry.
Explicitly models temporal and shape discontinuities at azimuth borders.
Integrates physically based rendering into 3DGS to enhance nighttime scene reconstruction accuracy
Introduces a plug-and-play mechanism to inject cross-view 3D geometric features into vision-language models
Formal specifications in temporal logic (LTLf) are systematically used to generate test scenarios.
NASA developed a method to distinguish lighting types at the pixel level in nighttime Earth photographs.
It separates design and real-time scheduling, optimizing each phase independently.