Data Efficient Human Intention Prediction: Leveraging Neural Network Verification and Expert Guidance
Ruixuan Liu Robotics Institute Carnegie Mellon University Pittsburgh, PA 15213、Changliu Liu Robotics Institute Carnegie Mellon University Pittsburgh, PA 15213 This work is in part supported by Ford Motor Company.
Abstract
Predicting human intention is critical to facilitating safe and efficient human-robot collaboration (HRC). However, it is challenging to build data-driven models for human intention prediction. One major challenge is due to the diversity and noise in human motion data. It is expensive to collect a massive motion dataset that comprehensively covers all possible scenarios, which leads to the scarcity of human motion data in certain scenarios, and therefore, causes difficulties in constructing robust and reliable intention predictors. To address the challenge, this paper proposes an iterative adversarial data augmentation (IADA) framework to learn neural network models from an insufficient amount of training data. The method uses neural network verification to identify the most “confusing” input samples and leverages expert guidance to safely and iteratively augment the training data with these samples. The proposed framework is applied to collected human datasets. The experiments demonstrate that our method can achieve more robust and accurate prediction performance compared to existing training methods.
中文速览
人机协作场景中，机器人需要准确预测人的动作意图才能保证安全与效率，但由于人体动作多样、传感器噪声等原因，相关训练数据往往既稀少又分布不均，导致神经网络模型泛化能力差。为此，研究者提出了一种"迭代对抗数据增强"（Iterative Adversarial Data Augmentation，IADA）框架：利用神经网络验证技术（neural network verification）自动找到当前模型最"困惑"的边界样本（即对抗样本），再由领域专家对这些样本进行标注，然后将其加回训练集并重新训练，如此循环迭代，逐步弥补数据不足的缺陷。在两项人体上肢动作意图预测任务上的实验表明，与现有训练方法相比，IADA 能将模型测试准确率提升超过 10%。这项工作的意义在于，它为数据稀缺的安全关键型人机协作场景提供了一种可控、可解释的数据增强思路，避免了 GAN 等生成式方法可解释性差的隐患，同时将人类专家的监督嵌入训练循环，从源头保障了扩充数据的质量与安全性。
原文 arXiv:2108.06871;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2108.06871v3