CLAIRE: Enabling Continual Learning for Real-time Autonomous Driving with a Dual-head Architecture
CLAIRE: Enabling Continual Learning for Real-time Autonomous Driving with a Dual-head Architecture
复制标题
DOI：
10.1109/isorc52572.2022.9812816
复制
发表时间：
2022-05
期刊：
影响因子：
--
通讯作者：
Hao Zhang-
中科院分区：
文献类型：
--
作者：
Hao Zhang-
自动驾驶车辆依靠预先训练的物体探测器来感知周围环境。然而，当遇到以前从未见过的场景时，迟到的决策可能会因感知到威胁而导致紧急制动。导致这种情况的图像序列提供了随着时间的推移学习和改进的潜力。然而，考虑到计算、存储和功率的限制，使用所有先前的训练数据进行即时重新训练是不可行的。更重要的是，已知将预训练的 CNN 仅暴露于新场景的图像会导致对已学特征的“灾难性遗忘”。这项工作做出了多项贡献：提出了一种新颖的轻量级双头检测网络架构来克服遗忘并支持对少量新图像进行快速车载持续学习，并评估自动驾驶持续学习方法的可行性。对我们的双头技术持续学习的图像的质量和数量进行了敏感性研究，包括对其实时适用性的评估。实验表明，与最先进的持续学习框架相比，我们的方法的准确性提高了 13%，性能提高了 5.8 倍。这使得它适合有实时约束的自动驾驶场景。源代码可通过 Github 获取。
Autonomous vehicles rely on a pre-trained object detector to perceive surroundings. However, when never seen before scenarios are encountered, late decisions may result in hard braking due to perceived threats. Image sequences leading to such a situation provide the potential to learn and improve over time. Yet instant re-training on board with all prior training data is infeasible given computational, storage and power constraints. What’s more, exposure of a pre-trained CNN to only images of the new scenario is known to result in “catastrophic forgetting” for already learned features.This work makes several contributions: A novel lightweight dual-head detection network architecture is proposed to overcome forgetting and to support fast on-board continual learning on small sets of new images and assesses the feasibility of continual learning methods for autonomous driving. A sensitivity study on the quality and quantity of continually learned images for our dual-head technique is performed, including an assessment of its real-time suitability. Experiments show that our method’s accuracy is improved by up to 13% and performance increases by 5.8X over a state-of-the-art continual learning framework. This makes it suitable for autonomous driving scenarios with real-time constraints. Source code is made available via Github.