AUGUstus: Autonomous Driving VLA Model with Unified Generation and Understanding
Vision-language-action (VLA) models show strong potential for end-to-end autonomous driving by leveraging world knowledge and understanding to excel in long-tail scenarios.
ProofPaper ↗
Key points
- Autoregressive methods sequentially generate chain-of-thought (CoT) and action tokens, fully leveraging VLM understanding, but high latency hinders deployment.
- To enable low-latency deployment while using CoT to assist action generation, we propose AUGUstus, an Autonomous-driving VLA model with Unified Generation and Understanding.
- We unify diffusion-based action generation and autoregressive CoT understanding in one model, where understanding only serves generation.
- AUGUstus-GRPO, a unified GRPO algorithm, jointly optimizes action and CoT, improving action performance.
Sources (2)
- [1]AUGUstus: Autonomous Driving VLA Model with Unified Generation and UnderstandingarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 12, 03:42 PM
Vision-language-action (VLA) models show strong potential for end-to-end autonomous driving by leveraging world knowledge and understanding to excel in long-tail scenarios.
Autoregressive methods sequentially generate chain-of-thought (CoT) and action tokens, fully leveraging VLM understanding, but high latency hinders deployment.
- [2]AUGUstus: Autonomous Driving VLA Model with Unified Generation and UnderstandingHugging Face Daily Papers · Oct 12, 12:00 AM · same content
Extractive summary: sentences quoted from the sources.
Before this
- Oct 11, 2026PhySimCode: A Benchmark and Evaluation Method for Physics Video to Code Generation
- Oct 11, 2026A Persian Vision-Language-Action Dataset and Benchmark for Ambiguity- and Infeasibility-Aware Instruction-to-Plan Grounding
- Oct 11, 2026Stereo-Planner: Learning Stereo-to-Trajectory Planning for Zero-Shot Sim-to-Real Quadrotor Navigation
- Oct 10, 2026Structural Chain-of-Thought: Reasoning with Executable Computational Graphs
- Oct 10, 2026Red-Teaming of Vision-Language-Action Models via Quality Diversity Environment Generation
- Oct 10, 2026KV4-RL: Unlocking Faster Asynchronous Rollouts for RLVR with INT4 KV Cache Quantization
