When Autonomous Systems Meet Accuracy and Transferability through AI: A Survey

Abstract

With widespread applications of artificial intelligence (AI), the capabilities of the perception, understanding, decision-making, and control for autonomous systems have improved significantly in recent years. When autonomous systems consider the performance of accuracy and transferability, several AI methods, such as adversarial learning, reinforcement learning (RL), and meta-learning, show their powerful performance. Here, we review the learning-based approaches in autonomous systems from the perspectives of accuracy and transferability. Accuracy means that a well-trained model shows good results during the testing phase, in which the testing set shares a same task or a data distribution with the training set. Transferability means that when a well-trained model is transferred to other testing domains, the accuracy is still good. Firstly, we introduce some basic concepts of transfer learning and then present some preliminaries of adversarial learning, RL, and meta-learning. Secondly, we focus on reviewing the accuracy or transferability or both of these approaches to show the advantages of adversarial learning, such as generative adversarial networks, in typical computer vision tasks in autonomous systems, including image style transfer, image super-resolution, image deblurring/dehazing/rain removal, semantic segmentation, depth estimation, pedestrian detection, and person re-identification. We furthermore review the performance of RL and meta-learning from the aspects of accuracy or transferability or both of them in autonomous systems, involving pedestrian tracking, robot navigation, and robotic manipulation. Finally, we discuss several challenges and future topics for the use of adversarial learning, RL, and meta-learning in autonomous systems.

Developing an Explainable Machine Learning-Based Personalised Dementia Risk Prediction Model: A Transfer Learning Approach With Ensemble Learning Algorithms

“…Finally, we employed a method proposed by Taylor and Stone ( 2009 ) to examine the efficacy of our transfer learning approach based on a learning ratio as expressed in Equation (4).…”

Unsupervised Monocular Depth Estimation in Highly Complex Environments

“…Domain adaptation. In monocular depth estimation, domain adaptation algorithms are mainly applied to adapt the model trained on synthetic datasets to real-world datasets [20], [35]- [37]. Compared with the ground truth obtained by different sensors in the real-world, the ground truth obtained from virtual environments is cheaper and easier.…”

“…The problem of estimating monocular depth in such varying environments is challenging but practical and important. Meanwhile, the perception of changing and complex environments is crucial for autonomous systems [19], [20], like robots and autonomous driving cars, and this problem has only received some initial attention [15], [18].…”

Quadrotor navigation in dynamic environments with deep reinforcement learning

“…On one hand, the performance of deep reinforcement learning methods depends critically on whether training data is sufficient (Tang et al, 2020). Especially, large amounts of training data are essential for the accuracy of perception and decision-making, while they will increase the cost of time and computing resources to train the networks (Zhang et al, 2020;Peng et al, 2002). Besides, dynamic obstacles in environments will make the training process more difficult to converge, which leads to a longer training phase.…”