MAVERIC: A Data-Driven Approach to Personalized Autonomous Driving
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper introduces MAVERIC (Manipulating Autonomous Vehicle Embedding Region for Individuals’ Comfort), a data-driven framework designed to personalize autonomous vehicle (AV) driving styles to increase user trust and acceptance. The authors address the limitation of prior approaches that either mimic user driving styles exactly or tune aggression without considering the user’s baseline style. They hypothesize that optimal AV behavior depends on both the user’s driving style and subjective factors like personality. MAVERIC learns a high-level model via a neural network that predicts personalized control parameters for low-level controllers, simultaneously generating a personalized embedding that represents an individual’s driving style. This embedding allows the system to modulate the level of aggression while maintaining other characteristics, such as headway distance. The methodology employs a network architecture with five subnetworks: Following Distance Predictor, Lane Change Predictor, Velocity Predictor, Style Predictor, and a Mutual Information module. These components are trained to minimize specific loss functions, ensuring the learned embedding captures salient driving style information. The system uses low-level controllers (Stanley, PI) to execute safe maneuvers based on the neural network’s high-level predictions. To modulate aggression, the framework shifts the personalized embedding along the gradient of a learned aggression dimension, derived from the Aggressive Driving Behavior (ADB) scale. The approach was validated through two human-subject studies using a high-fidelity 6-DOF driving simulator. Study 1 involved 30 participants to train the model, while Study 2 involved 24 participants to test the system under four conditions: Mimic (matching user style), Aggressive, Cautious, and Perpendicular (constant aggression, altered other traits). Results demonstrate that MAVERIC effectively mimics end-user driving styles and successfully modulates aggression. Participants rated the Mimic condition as significantly more similar to their own aggressiveness level ($p = .002$) compared to other conditions. The study further investigated homophily, the preference for driving styles similar to one’s own. The authors found that personality traits ($p < .001$), perceived similarity ($p < .001$), and high-velocity driving style ($p = .0031$) significantly modulate the effect of homophily. This indicates that user preference for AV behavior is not solely determined by mimicry but is influenced by subjective factors and specific driving contexts. The significance of this work lies in providing a robust method for personalizing AVs that accounts for both objective driving metrics and subjective user characteristics. By demonstrating that AVs can be tuned to be more or less aggressive than the user while maintaining safety and other style attributes, the study offers a pathway to optimize user acceptance. The findings suggest that future AV personalization strategies must integrate user personality and perceived similarity to effectively tailor driving behaviors, moving beyond simple mimicry or generic tuning.
Key finding
The MAVERIC framework successfully personalized autonomous vehicle driving styles to mimic user preferences and modulate aggression, with user acceptance significantly influenced by personality, perceived similarity, and driving style.
Methodology
simulator
Sample size: 54
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via author_sweep_intake on 2026-05-28.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | author_sweep | — | — | 2 | 2026-05-28 |
| archive | success | canonical_url | — | — | 1 | 2026-06-06 |
| extract | success | cached | — | — | 3 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-04 |
| chunk | success | chunk | — | — | 1 | 2026-06-04 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-04 |
| enrich | success | — | — | — | 1 | 2026-05-28 |
| promote | success | — | — | — | 1 | 2026-06-04 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 2 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 15 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Theoretical Contribution: computational model