Paper Detail

PRIME: Perception Feedback with Situational Memory Embeddings in VLA Models

Erik Deinzer, Naya Baslan, Luca Paparusso, Narunas Vaskevicius, Peter Knott, Luigi Palmieri

Browse

Workflow Queues

arxiv Score 18.8

Published 2026-09-18 · First seen 2026-09-21

General AI

Open paper source

Abstract

Current Vision-Language-Action (VLA) models for autonomous driving operate primarily through feedforward inference across the perception--reasoning--planning hierarchy. While modern architectures maintain temporal recurrence within the perceptual module, early perception remains blind to downstream reasoning and navigation goals, processing visual inputs agnostically without prioritizing cues informed by prior decisions. To bridge this gap, this paper introduces PRIME, a learned feedback mechanism that conditions the VLA perceptual queries on a novel Situational Memory. By aggregating latent representations of past perception, reasoning, navigation goals, and predicted behaviors across an L-step window via cross-attention, PRIME enables intent-driven perceptual attention at minimal computational cost, adding only a maximum of 29.7M parameters (0.41% of the 7.3B-parameter base model). Evaluated on the Bench2Drive closed-loop benchmark, PRIME achieves a state-of-the-art Driving Score of 82.47 (+4.73 over ORION) and a Success Rate of 60.00% (+5.38 percentage points), the highest reported Driving Score among published VLAs trained on Think2Drive demonstrations.

Workflow Status

Review status
pending
Role
unreviewed
Read priority
now
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

Similar Papers

FM-VLA: Force-based Memory for Vision-Language-Action Models in Contact-Rich Manipulation 2026-07-20 · Score 9.3 General AI Lightweight LLM Agent Memory with Small Language Models 2026-04-09 · Score 12.3 Research Track A · General AI Improving Vision-language Models with Perception-centric Process Reward Models 2026-04-27 · Score 16.5 General AI Toward Secure and Reliable PDDL Formalization of Large Language Models with Planner-in-the-Loop Feedback 2026-06-29 · Score 12.3 General AI Beyond Perception Errors: Semantic Fixation in Large Vision-Language Models 2026-04-13 · Score 5.5 General AI

BibTeX

@article{deinzer2026prime,
  title = {PRIME: Perception Feedback with Situational Memory Embeddings in VLA Models},
  author = {Erik Deinzer and Naya Baslan and Luca Paparusso and Narunas Vaskevicius and Peter Knott and Luigi Palmieri},
  year = {2026},
  abstract = {Current Vision-Language-Action (VLA) models for autonomous driving operate primarily through feedforward inference across the perception--reasoning--planning hierarchy. While modern architectures maintain temporal recurrence within the perceptual module, early perception remains blind to downstream reasoning and navigation goals, processing visual inputs agnostically without prioritizing cues informed by prior decisions. To bridge this gap, this paper introduces PRIME, a learned feedback mechani},
  url = {https://arxiv.org/abs/2609.22040},
  keywords = {cs.CV, cs.RO},
  eprint = {2609.22040},
  archiveprefix = {arXiv},
}

Metadata

{}