デルフト工科大学、自動運転車の社会的走行を支援する新モデルを開発(Merging like a human: TU Delft develops new model to help self-driving cars drive socially)

2024-11-04 オランダ・デルフト工科大学（TUDelft)

デルフト工科大学（TU Delft）の研究者は、自己中心的か協調的かといった人間の運転者の社会的行動を評価し、それに基づいて自動運転車が他の車両の行動を予測し、より安全に運転できる新しいモデルを開発しました。このモデルは、車線変更や合流などの状況で他の車両の行動を25%精度向上させることが確認されています。この技術は、自動運転車が人間の運転者とより自然に相互作用し、交通の安全性と効率性を高めることを目指しています。

＜関連情報＞

ダイアド合流の相互作用モデルにより、制御入力から意思決定までの人間のドライバーの行動が説明できる A model of dyadic merging interactions explains human drivers’ behavior from control inputs to decisions

Olger Siebinga, Arkady Zgonnikov, David A Abbink
PNAS Nexus  Published:24 September 2024
DOI:https://doi.org/10.1093/pnasnexus/pgae420

Models of highway interactions and the aspects of interaction they describe in a merging scenario. A) a typical interactive merging scenario taken from the HighD dataset (5) (dataset 60, vehicles 458 and 468). In this scenario, the driver of the green vehicle wants to merge onto the highway. This vehicle has a position advantage but a significantly lower velocity compared to the purple vehicle driving on the highway. B) the vehicles’ position traces, the gap between the vehicles, and the individual velocity traces. In this example, the joint high-level decision is indicated at I: Green merges ahead of Purple (i.e. the green, merging vehicle goes first). Both vehicles individually contribute to this decision by accelerating and decelerating respectively (at 1). After the decision has been made, Green keeps accelerating and thereby individually contributes to maintaining a safety margin while purple stops decelerating (2). The gap between the vehicles, when Green crosses the lane marker (II), denotes the joint safety margin. Finally, the underlying characteristics of the individual vehicle control inputs are depicted here as the total velocity traces (3). We evaluate velocity traces instead of raw accelerations because they are easier to perceive for other drivers and provide more insight into the trend of the driver’s actions since they are less noisy. These individual and joint perspectives on the three levels of behavior are also indicated in (C). C) the three levels of behavior in between Michon’s operational and tactical behavior (6). It also shows five modeling strategies for merging interactions, each with examples from the literature. The icons indicate if the models describe a single decision at the start of an interaction (one-shot), repeated decisions (multishot), or continuous behavior. Every modeling strategy captures part of the overall interactive behavior, but none covers all five aspects. We postulate that a model capturing all three levels of individual and joint behavior simultaneously is likely to have captured the underlying mechanisms of merging behavior.

Abstract

Safe and socially acceptable interactions with human-driven vehicles are a major challenge in automated driving. A good understanding of the underlying principles of such traffic interactions could help address this challenge. Particularly, accurate driver models could be used to inform automated vehicles in interactions. These interactions entail complex dynamic joint behaviors composed of individual driver contributions in terms of high-level decisions, safety margins, and low-level control inputs. Existing driver models typically focus on one of these aspects, limiting our understanding of the underlying principles of traffic interactions. Here, we present a Communication-Enabled Interaction model based on risk perception, that does not assume humans are rational and explicitly accounts for communication between drivers. Our model can explain and reproduce observed human interactions in a simplified merging scenario on all three levels. Thereby improving our understanding of the underlying mechanisms of human traffic interactions and posing a step towards interaction-aware automated driving.

0108交通物流機械及び建設機械
人間の社会的行動評価他車両行動予測自動運転車
ad
ad
Share
Follow
ad
テック・アイ技術情報研究所
ad
タイトルとURLをコピーしました