From Prompts to Pavement Through Time: Temporal Grounding in Agentic Scene-to-Plan Reasoning
The article discusses a study on temporal grounding in scene-to-plan reasoning for autonomous vehicles. It highlights the limitations of current models that treat time as a secondary property, which affects reasoning consistency. The research introduces new planner architectures and evaluates their performance, revealing insights into predictive hazard reasoning and the challenges of prompt-based temporal grounding.
- ▪The study focuses on enhancing scene interpretation and planning in autonomous vehicles using large language and multimodal models.
- ▪It introduces three planner architectures with varying levels of temporal integration to assess their effectiveness.
- ▪Qualitative analysis indicates improvements in predictive hazard reasoning and strategic divergence, despite no significant gains in standard correctness metrics.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.19824 (cs) [Submitted on 19 May 2026] Title:From Prompts to Pavement Through Time: Temporal Grounding in Agentic Scene-to-Plan Reasoning Authors:Ahmed Y. Gado, Omar Y. Goba, Alaa Hassanein, Catherine M. Elias, Ahmed Hussein View a PDF of the paper titled From Prompts to Pavement Through Time: Temporal Grounding in Agentic Scene-to-Plan Reasoning, by Ahmed Y. Gado and 4 other authors View PDF HTML (experimental) Abstract:Recent attempts to support high-level scene interpretation and planning in Autonomous Vehicles (AVs) using ensembles of Large Language Models (LLMs) and Large Multimodal Models (LMMs) continue to treat time as a secondary property.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
