The Ethical Decision Head: Operationalizing Normative Ethics in Autonomous Vehicles via Reinforcement Learning from Human Feedback

📄 arXiv: 2608.16710v1 📥 PDF

作者: Thomas Mbrice, Ammar Ali, Sami Mian, Khai Hern Low, Eric Chen, Arshia Aghajani, Wolf Schäfer, Amin Shirangi

分类: cs.LG

发布日期: 2026-08-17


💡 一句话要点

提出伦理决策头以解决自主车辆道德决策问题

🎯 匹配领域: 支柱一:机器人控制 (Robot Control) 支柱二:RL算法与架构 (RL & Architecture)

关键词: 伦理决策 自主车辆 深度强化学习 功利主义 康德伦理 人类反馈 道德推理

📋 核心要点

  1. 现有自主车辆决策系统在处理道德决策时面临挑战,缺乏有效的伦理框架支持。
  2. 论文提出伦理决策头(EDH),通过深度强化学习将伦理推理转化为可微分的奖励信号,促进道德行为学习。
  3. 实验结果显示,康德框架提供了训练稳定性,而功利主义代理则反映了人类对自我牺牲的偏好,揭示了伦理学习的复杂性。

📝 摘要(中文)

随着自主车辆(AVs)接近4级和5级操作能力,其决策系统不仅需处理安全问题,还需考虑道德权重。本文详细介绍了伦理决策头(EDH),这是一个深度强化学习框架,通过可微分的奖励信号编码伦理推理,使得策略梯度代理能够在与CARLA仿真环境对齐的场景中学习道德驱动行为。本文实例化并评估了两种规范性框架:一种是最小化总伤亡的功利主义框架,另一种是作为绝对命令的康德框架。EDH通过近端策略优化(PPO)进行训练,结果显示在人工监督下,规范伦理框架的可学习性存在不对称性,功利主义代理学习到了人类在实践中对自我牺牲的偏好,揭示了伦理学习的复杂性。

🔬 方法详解

问题定义:本文旨在解决自主车辆在道德决策中的不足,现有方法未能有效整合伦理框架,导致决策缺乏人性化考虑。

核心思路:提出伦理决策头(EDH),通过深度强化学习将伦理推理转化为可微分的奖励信号,使得代理能够在复杂场景中学习道德驱动的行为。

技术框架:整体架构包括两个主要模块:伦理决策头和近端策略优化(PPO)。伦理决策头负责将伦理框架转化为奖励信号,PPO则用于优化策略。

关键创新:最重要的技术创新在于将伦理推理与深度强化学习相结合,形成了一个新的学习范式,使得代理能够在道德决策中学习人类的真实偏好。

关键设计:采用Bradley-Terry奖励模型,通过200个即将发生碰撞的场景进行训练,设计了功利主义和康德两种框架,确保了训练的稳定性和有效性。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

实验结果显示,康德框架在训练中提供了稳定性,而功利主义代理则意外地学习到了人类对自我牺牲的偏好。这表明,伦理学习并非如哲学家所定义,而是反映了人类的实际行为。此发现对伦理AI的设计具有重要启示。

🎯 应用场景

该研究的潜在应用领域包括自主驾驶汽车、智能交通系统和机器人伦理决策等。通过将伦理决策融入自主系统,可以提高其在复杂环境中的决策能力,确保更安全和更符合人类价值观的操作。未来,这一方法可能推动更广泛的伦理AI研究和应用。

📄 摘要(原文)

As autonomous vehicles (AVs) approach Level 4 and Level 5 operational capability [SAE International, 2018], their on- board decision systems must handle not only safety-critical locomotion but also their subsequent moral weight. This paper details the Ethical Decision Head (EDH), a deep re- inforcement learning (RL) framework that encodes ethical reasoning as a differentiable reward signal, enabling a pol- icy gradient agent to learn morally-aligned driving behavior in scenarios whose state representation is aligned with the CARLA simulation environment [Dosovitskiy et al., 2017]. Two normative frameworks are instantiated and evaluated: a Utilitarian framework minimizing total casualties and a Kan- tian framework enforcing course maintenance as a categori- cal imperative. The EDH is trained via Proximal Policy Op- timization (PPO) [Schulman et al., 2017] against a Bradley- Terry reward model [Bradley and Terry, 1952] learned from pairwise human preference annotations over 200 collision- imminent scenarios. Results reveal an asymmetry in the learnability of normative ethical frameworks under human su- pervision. The Kantian condition, which reduces to a con- stant prediction task under the codebook, serves as a pipeline control: it confirms training stability and rules out infrastruc- ture failure as an explanation for the utilitarian result. The Utilitarian agent learned something more unsettling: human raters rewarded self-sacrifice over casualty minimization, and the model learned that preference faithfully. This divergence between what humans prescribe in theory and what they re- ward in practice suggests that RLHF does not learn ethics as philosophers define it, but as humans live it.