Carla
PulseAugur coverage of Carla — every cluster mentioning Carla across labs, papers, and developer communities, ranked by signal.
10 day(s) with sentiment data
-
New imitation learning policy enables autonomous driving in CARLA simulator
Researchers have developed a new imitation learning approach for autonomous driving within the CARLA simulator. This method trains a compact, multimodal policy using historical data including RGB images, LiDAR, vehicle …
-
GrayTrack system enhances vehicle tracking using indirect sensor data
Researchers have developed GrayTrack, a novel system designed to improve vehicle tracking by integrating indirect observations from third-party sensors. This approach addresses limitations in direct sensing, such as pri…
-
New graph-based method evaluates LiDAR simulation fidelity for autonomous driving
Researchers have developed a new graph-based framework to assess the structural fidelity of simulated LiDAR point clouds, which are crucial for validating autonomous driving systems. This method goes beyond traditional …
-
Spiking Neural Network Achieves High Accuracy in Pedestrian Crossing Intent Classification
Researchers have developed a novel convolutional spiking neural network (Conv-SNN) for classifying pedestrian crossing intent using event-based vision. This approach converts real-world driving footage into synthetic dy…
-
Formal verification method proves AI steering failures without simulation
Researchers have developed a formal verification method called bound propagation to test AI-based automated driving systems. This technique analyzes trained neural network weights to predict potential steering failures …
-
New method evaluates AI-generated maps for autonomous driving
Researchers have developed a new method for evaluating Bird's-Eye View (BEV) maps generated by Cross-View Transformers (CVTs) for autonomous driving. These BEV maps are crucial inputs for behavioral cloning policies. Th…
-
Connected vehicles use AI for situation awareness to optimize data distribution
Researchers have developed a novel approach for intelligent data distribution in connected vehicles by enhancing situation awareness. This method uses Bird's-Eye-View images, object detection, and semantic segmentation …
-
Autonomous driving shifts to unified VLA models, challenging modular stacks
New Vision-Language-Action (VLA) models are emerging that aim to unify perception, reasoning, and control in autonomous driving, moving away from traditional modular stacks. Three prominent models—AutoVLA from UCLA, NVI…
-
LLM framework CoLMIN enhances autonomous driving negotiation
Researchers have developed CoLMIN, a new framework for cooperative autonomous driving that utilizes Large Language Models (LLMs) to improve decision-making and consensus among vehicles. Unlike previous methods that coul…
-
DiDrive framework enhances autonomous driving safety with diffusion models
Researchers have introduced DiDrive, a novel framework designed to enhance the safety and reliability of autonomous driving systems using offline reinforcement learning. The framework incorporates a Risk-Aware Hierarchi…
-
New attack exploits autonomous vehicle sensor calibration
Researchers have developed an Adversarial Calibration Attack (ACA) targeting the online calibration systems of autonomous vehicles. This attack exploits the process by which vehicles detect and correct sensor misalignme…
-
Self-play driving policies show mixed results on real-world maps
Researchers have explored the effectiveness of self-play for training autonomous driving policies, extending previous work with models like Gigaflow and Puffer-Drive. By shifting from MLPs to Transformers and training o…
-
New SynCrash pipeline enables zero-shot accident detection in traffic videos
Researchers have developed SynCrash, a novel three-stage pipeline designed for zero-shot accident detection and localization in traffic surveillance videos. This system addresses the CVPR 2026 Challenge by identifying a…
-
New AI method trains self-driving perception without manual labels
Researchers have proposed a novel approach to train 3D perception systems for self-driving cars without manual annotation. This method, termed "infrastructure-taught, label-free 3D perception," utilizes roadside units (…
-
New 3D adversarial attack targets autonomous vehicle perception systems
Researchers have developed a new adversarial attack method called 3DGAA, designed to realistically and robustly target camera-based perception systems in autonomous vehicles. This framework generates view-consistent, ge…
-
New GS-Net module boosts autonomous vehicle data reuse with faster 3DGS
Researchers have developed GS-Net, a novel module designed to enhance data reuse across different autonomous vehicles. This plug-and-play system aggregates geometric context from sparse Structure-from-Motion point cloud…
-
New HDR imaging method DMEB introduced for robotic vision systems
Researchers have introduced a new method called Depth-guided Multi-view Exposure Bracketing (DMEB) for capturing high dynamic range (HDR) images in robotic vision systems. This approach addresses the challenge of reliab…
-
Autonomous driving system uses risk-adaptive edge-cloud for efficient communication
Researchers have developed a novel risk-adaptive edge-cloud architecture designed to improve communication efficiency and reduce latency in autonomous driving systems. This architecture utilizes an onboard vision-langua…
-
New AI framework teaches autonomous vehicles ethical decision-making
Researchers have developed an "Ethical Decision Head" (EDH) framework using deep reinforcement learning to imbue autonomous vehicles with ethical reasoning capabilities. The EDH framework encodes ethical principles as a…
-
New framework evaluates autonomous driving VLA models across domains
Researchers have developed SSP, an evaluation framework for autonomous driving vision-language-action (VLA) models. SSP matches events across synthetic, simulated, and physical domains to accurately assess model perform…