2026-05-27原文

When Does Adaptive Guidance Help? Belief-Aware Privileged Distillation for Autonomous Driving Under Partial Observability

This paper proposes Belief-Aware GSAC (BA-GSAC), which adaptively modulates the distillation coefficient λ via ensemble disagreement, and systematically investigates when adaptive guidance is beneficial for autonomous driving under partial observability. Experiments show benefits under mild to moderate occlusion, but under severe occlusion the adaptive coefficient collapses due to 'observability blindness'—the ensemble predicts partial observations and fails to detect missing information. Proposed fix: train ensemble on full-state predictions. Simple linear decay schedule outperforms adaptive methods, indicating stability gain stems from scheduling effect.

Article intelligence

EngineersAdvanced

Key points

BA-GSAC dynamically adjusts distillation coefficient using ensemble disagreement for knowledge distillation in autonomous driving.
Adaptive guidance helps under mild to moderate partial observability but fails under severe occlusion due to observability blindness.
Simple linear decay schedule achieves best performance under severe POMDP, with stability benefits driven by scheduling rather than adaptivity.
Training ensemble on full-state predictions is recommended to improve uncertainty awareness.

Why it matters

This matters because BA-GSAC dynamically adjusts distillation coefficient using ensemble disagreement for knowledge distillation in autonomous driving.

Technical impact

May affect model selection, inference cost, product capability, and evaluation benchmarks.

[2605.26155] When Does Adaptive Guidance Help? Belief-Aware Privileged Distillation for Autonomous Driving Under Partial Observability

[Submitted on 24 May 2026]

Title:When Does Adaptive Guidance Help? Belief-Aware Privileged Distillation for Autonomous Driving Under Partial Observability

View a PDF of the paper titled When Does Adaptive Guidance Help? Belief-Aware Privileged Distillation for Autonomous Driving Under Partial Observability, by Mehmet Haklidir

View PDF HTML (experimental)

Abstract:Guided Soft Actor-Critic (GSAC) distills knowledge from a privileged full-state teacher to a partial-observation student for autonomous driving, but uses a fixed distillation coefficient lambda regardless of the agent's uncertainty. We present Belief-Aware GSAC (BA-GSAC), which modulates lambda via ensemble disagreement, and use it as a testbed for a systematic empirical study asking: when does adaptive guidance actually help? Evaluating five strategies (fixed lambda in {0.01, 0.1}, adaptive, linear decay, and vanilla SAC) across three POMDP difficulty levels on Highway-Env, we find that preliminary single-seed runs suggest benefits under mild and moderate partial observability, but under severe occlusion (evaluated with 3 seeds for all methods) the adaptive coefficient collapses to lambda_min within about 3K steps. We trace this to an observability blindness phenomenon: because the ensemble predicts partial observations, it achieves low disagreement even under heavy occlusion, modeling what is visible but unable to detect what is missing. We diagnose the root cause and propose an architectural fix (training the ensemble on full-state predictions using the guiding actor's privileged access); while not validated here, we show that even with current limitations, the warmup phase provides measurable stabilization (CV=13.3% vs. 29.8% for constant lambda=0.01). In fact, a simple deterministic linear decay schedule achieves the best severe-POMDP performance across all metrics (mean 116.5, CV=8.9%), suggesting that the scheduling effect, not the ensemble, drives the stability benefit. These findings provide practical guidance for designing uncertainty-aware teacher-student frameworks and highlight ensemble prediction targets as an important design choice.

Comments: 9 pages, 3 figures, 7 tables. Accepted at CVPR 2026 Workshop on Autonomous Driving (WAD)

Subjects:

Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)

ACM classes: I.2.9; I.2.6

Cite as: arXiv:2605.26155 [cs.RO]

(or arXiv:2605.26155v1 [cs.RO] for this version)

https://doi.org/10.48550/arXiv.2605.26155

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Mehmet Haklidir [view email] [v1] Sun, 24 May 2026 04:41:30 UTC (1,079 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled When Does Adaptive Guidance Help? Belief-Aware Privileged Distillation for Autonomous Driving Under Partial Observability, by Mehmet Haklidir

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.RO

new | recent | 2026-05

Change to browse by:

cs cs.AI cs.LG

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)