Tag
Tesla FSD Supervised reportedly avoided a crash by reacting 0.17 seconds before the at-fault vehicle moved, demonstrating superhuman reflexes and 360-degree awareness.
This paper proposes Dreamer-SAC, a model-based reinforcement learning framework that integrates a recurrent state-space world model with soft actor-critic in latent space for sample-efficient autonomous driving. It outperforms DreamerV3, SAC, and PPO baselines while requiring fewer real environment interactions.
This paper proposes Threat-guided Policy-aware Scene Perturbation (TPSP), a method that augments online reinforcement learning for safe autonomous driving by perturbing scenes in a policy-aware, targeted manner to generate high-value safety-critical experiences. Experiments on NAVSIM v2 show improved safety learning efficiency with about 4 million kilometers of simulated driving data.
Tesla shares the story of Justin Strafuss, a driver with cerebral palsy and limited peripheral vision, who says FSD Supervised restores his driving freedom and reduces the mental load of manual driving.
A California reporter explains why he traded his Lincoln Navigator for a Tesla Model Y, citing Full Self-Driving as the primary reason.
Kodiak Robotics announces its autonomous driving solution is now deployed in 35 customer-owned trucks operating without humans, the largest such fleet, and will ring the Nasdaq opening bell for its one-year public milestone.
Tesla promotes its FSD Supervised feature, claiming that being driven by the car is the ultimate luxury.
Introduces CMU-Drive, a closed-loop benchmark for cooperative multi-agent autonomous driving, and V2V-VLA, a vision-language-action model that jointly generates driving actions, waypoints, reasoning, and communication policies. This provides the first benchmark and baseline for cooperative VLA driving.
Tesla celebrates the first owner to complete 25,000 miles on the FSD Supervised streak counter without touching the wheel, highlighting the system's reliability.
San Francisco 49ers coach Kyle Shanahan revealed that his Tesla was on Autopilot during a recent crash in Palo Alto, while taking responsibility and cautioning that drivers must stay engaged.
Sawyer Merritt congratulates David Moss as the first Tesla owner to reach 25,000 miles on the FSD streak counter, a milestone equivalent to one lap around Earth.
Tesla teases a before-and-after comparison of life with FSD Supervised, hinting at the transformative impact of its supervised full self-driving feature.
Elon Musk shares Paul Graham's anecdote about Jessica buying a Tesla and being impressed by its self-driving capabilities, highlighting the automaker's ability to inspire customers.
Tesla Europe announces FSD Supervised has been trained and tested on 2.2 million km across 19 EU countries, handling rare edge cases.
INTraJ is a unified framework for trajectory prediction that decomposes social influence into two stages: planning with future social information and local reaction from residuals, achieving state-of-the-art results on Argoverse 2, ETH/UCY, and SDD benchmarks.
SimWAM is a simple yet effective World Action Model for end-to-end autonomous driving that uses video generation purely as a training signal, achieving state-of-the-art 91.5 PDMS on NAVSIM while reducing inference latency.
Jensen Huang announced that NVIDIA has officially open-sourced its autonomous driving reasoning model Alpamayo 2 Super. The model can understand complex scenes and think before acting, and is suitable for Robotaxi, trucks, delivery vehicles, etc. It is open for commercial use under the OpenMDW-1.1 license.
This paper audits NAVSIM v2.2's defensive driving scoring, showing that a shared reference-conditioned forgiveness rule combined with numerical instability can cause actor-blind probes to outrank human replay, undermining the benchmark's validity.
The paper presents NSF-HRPT, a framework that combines a Neural Semantic Field with a Hierarchical Risk Perception Tree for quantitative risk assessment in safety-critical autonomous driving scenarios, achieving state-of-the-art performance on synthetic benchmarks and near-state-of-the-art results on real-world datasets.
Wayve announces GAIA-4, a multimodal world model powering closed-loop simulation for safety-critical evaluation of end-to-end autonomous driving models, enabling counterfactual replay of cyclist and pedestrian interactions.