Formalizing Human Machine Communication in the Context of Autonomous Vehicles

Gopalswamy, Swaminathan; Saripalli, Srikanth; Shell, Dylan; Hickman, Jeff; Hsu, Ya-Chuan · 2020 · ROSA P / Safety through Disruption (Safe-D) University Transportation Center (UTC)

archive: archived pipeline: cataloged verified

Get this paper ↗ (full text — opens at the source; we link to it, we don't host it)

Summary

This study addresses the critical safety challenge of integrating tacit human-pedestrian communication into autonomous vehicle (AV) decision-making. While current AVs rely on sensor data regarding pedestrian location and speed, they often fail to interpret implicit social cues, such as eye contact or gestures, which govern safe interactions in manual driving. The research aims to formalize these communication patterns into a computational model that allows AVs to mimic human-like, socially aware driving strategies. The methodology comprised two parallel tracks: a naturalistic field study and the development of a formal behavioral model. The field study utilized a 2008 Cadillac STS instrumented with a MiniDAS system to capture video, audio, and CAN bus data from 10 drivers across a 2x2x2 factorial design. Variables included driving context (vehicle marked as "self-driving" or not), route type (signalized vs. unsignalized crosswalks), and narration condition (think-aloud protocol). Simultaneously, the researchers developed a Domain Specific Language (DSL) using the Meta Programming System (MPS) to precisely describe communication scenarios. This DSL informed a probabilistic model based on a Partially Observable Markov Decision Process (POMDP) to formalize vehicle-pedestrian interactions, which was evaluated through simulations rather than real-time vehicle deployment due to project timeline constraints. Key findings from the field study revealed that while drivers altered visible behaviors—such as hand position and waving—when their vehicle was marked as autonomous, their actual yielding behavior remained statistically unchanged. Pedestrians also showed no significant behavioral changes in response to the autonomous vehicle sign. Regression analysis identified pedestrian distance to the crossing and eye contact (both driver and pedestrian) as the strongest predictors of driver yield behavior. The formal modeling component demonstrated that the POMDP framework could generate optimal action policies for AVs. Simulations showed the model could adapt to different pedestrian risk profiles, such as cautious versus reckless crossing behaviors, by adjusting vehicle strategies based on inferred intent. The significance of this work lies in its provision of a structured framework for translating implicit social communication into executable AV code. By identifying specific behavioral predictors like eye contact and distance, the study offers concrete variables for programming SAE Level 4 and 5 AVs to enhance safety and efficiency. The development of the DSL and POMDP model provides a foundation for future research to create richer communication languages and faster solvers for real-time operation, ultimately enabling AVs to engage in the nuanced, cooperative interactions necessary for safe shared roadways.

Key finding

A partially observable Markov decision process model formalized through a domain-specific language successfully generated autonomous vehicle policies that mimicked human driver-pedestrian interactions in simulation.

Methodology

mixed_methods

Sample size: 10

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 bulk_ingest_rosap on 2026-05-23 (6 acquisition events logged).

StageOutcomeToolModelPromptAttemptsCompleted
discover success rosap 2 2026-05-23
archive success 1 2026-05-23
extract success cached 2 2026-06-10
clean success 1 2026-06-01
chunk success 1 2026-06-01
embed success 1 2026-06-02
enrich success 1 2026-05-23
promote success 1 2026-05-23
summarize success llm qwen3.6-27b-prismaquant summ-v5 3 2026-06-10
tag success vector_similarity 19 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).