🛰️ Daily AI Frontier
‹ back to 2026-08-10

Depth-Wise Probing and Pruning of the Planning Token in a Driving Vision-Language-Action Model

Research Embodied AI & Robotics

Ranking

Overall 62
Content 70
Popularity 42

Observed public metrics from 1 member.

Merged summary

TL;DR - An arXiv analysis of a driving vision-language-action model (ORION on Bench2Drive) that probes the single "planning token" at each of 32 decoder layers, showing semantic intent emerges almost immediately while planner-compatible formatting accrues slowly with depth. It matters because it exposes redundant decoder depth in VLA stacks and enables ~1.33× inference speedup with modest accuracy loss.

  • Using the generative planner as a trajectory-space "logit lens," navigation-command probe accuracy hits 97.7% after just the first decoder layer (vs. 16.7% chance), indicating intent is linearly decodable early.
  • Compatibility with the frozen native planner improves only gradually, with open-loop Avg-L2 bottoming out at 2.11 m at the final layer; learned readouts from layer 1 recover much of that gap, suggesting a representation-format mismatch rather than missing information.
  • Ranking layers by the angular deviation they induce in the planning token allows pruning 8 of 32 layers for ~5% relative open-loop error increase and a measured 1.33× decoder speedup.
  • Scope is explicitly limited: results come from one ORION checkpoint and the Bench2Drive setup, and no family-specific degradation was statistically resolved at the evaluated sample size.

Sources (1)

Depth-Wise Probing and Pruning of the Planning Token in a Driving Vision-Language-Action Model

arXiv cs.RO Harisankar Babu, Benjamin Coors, Christopher Lang, Hendrik Berkemeyer, Tamim Asfour, Simon Foell 2026-08-07 arXiv:2608.07361
Public signals Hugging Face upvotes 0 · Semantic Scholar citations 0 · Semantic Scholar influential citations 0
Providers: Hugging Face · Upvotes 0 OpenAlex · N/A Publisher · N/A Semantic Scholar · Citations 0 · Influential citations 0 X · N/A Fetched 2026-09-09 08:18:10.300237 UTC

TL;DR - An arXiv analysis of a driving vision-language-action model (ORION on Bench2Drive) that probes the single "planning token" at each of 32 decoder layers, showing semantic intent emerges almost immediately while planner-compatible formatting accrues slowly with depth. It matters because it exposes redundant decoder depth in VLA stacks and enables ~1.33× inference speedup with modest accuracy loss.

  • Using the generative planner as a trajectory-space "logit lens," navigation-command probe accuracy hits 97.7% after just the first decoder layer (vs. 16.7% chance), indicating intent is linearly decodable early.
  • Compatibility with the frozen native planner improves only gradually, with open-loop Avg-L2 bottoming out at 2.11 m at the final layer; learned readouts from layer 1 recover much of that gap, suggesting a representation-format mismatch rather than missing information.
  • Ranking layers by the angular deviation they induce in the planning token allows pruning 8 of 32 layers for ~5% relative open-loop error increase and a measured 1.33× decoder speedup.
  • Scope is explicitly limited: results come from one ORION checkpoint and the Bench2Drive setup, and no family-specific degradation was statistically resolved at the evaluated sample size.
item →