Domain adaptation enables generalized learning in new environments by transferring knowledge from label-rich source domains to label-scarce target domains. As a more realistic extension, partial domain adaptation (PDA) relaxes the assumption of fully shared label space, and instead deals with the scenario where the target label space is a subset of the source label space. In this paper, we propose a Reinforced Adaptation Network (RAN) to address the challenging PDA problem. Specifically, a deep reinforcement learning model is proposed to learn source data selection policies. Meanwhile, a domain adaptation model is presented to simultaneously determine rewards and learn domain-invariant feature representations. By combining reinforcement learning and domain adaptation techniques, the proposed network alleviates negative transfer by automatically filtering out less relevant source data and promotes positive transfer by minimizing the distribution discrepancy across domains. Experiments on three benchmark datasets demonstrate that RAN consistently outperforms seventeen existing state-of-the-art methods by a large margin.

Multi-Source Domain Adaptation (MSDA) is a more practical domain adaptation scenario in real-world scenarios, which relaxes the assumption in conventional Unsupervised Domain Adaptation (UDA) that source data are sampled from a single domain and match a uniform data distribution. The MSDA is more challenging due to the existence of different domain shifts between distinct domain pairs. When considering videos, the negative transfer would be provoked by spatial-temporal features and can be formulated into a more challenging Multi-Source Video Domain Adaptation (MSVDA) problem. In this paper, we address the MSVDA problem by proposing a novel Temporal Attentive Moment Alignment Network (TAMAN) which aims for effective feature transfer by dynamically aligning both spatial and temporal feature moments. The TAMAN further constructs robust global temporal features by attending to dominant domain-invariant local temporal features with high local classification confidence and low disparity between global and local feature discrepancies. To facilitate future research on the MSVDA problem, we introduce comprehensive benchmarks, covering extensive MSVDA scenarios. Empirical results demonstrate a superior performance of the proposed TAMAN across multiple MSVDA benchmarks.

Despite the great potential of reinforcement learning (RL) in solving complex decision-making problems, generalization remains one of its key challenges, leading to difficulty in deploying learned RL policies to new environments. In this paper, we propose to improve the generalization of RL algorithms through fusing Self-supervised learning into Intrinsic Motivation (SIM). Specifically, SIM boosts representation learning through driving the cross-correlation matrix between the embeddings of augmented and non-augmented samples close to the identity matrix. This aims to increase the similarity between the embedding vectors of a sample and its augmented version while minimizing the redundancy between the components of these vectors. Meanwhile, the redundancy reduction based self-supervised loss is converted to an intrinsic reward to further improve generalization in RL via an auxiliary objective. As a general paradigm, SIM can be implemented on top of any RL algorithm. Extensive evaluations have been performed on a diversity of tasks. Experimental results demonstrate that SIM consistently outperforms the state-of-the-art methods and exhibits superior generalization capability and sample efficiency.

Real-time path planning is crucial for intelligent vehicles to achieve autonomous navigation. In this paper, we propose a novel deep neural network (DNN) based method for real-time online path planning in unknown cluttered environments. Firstly, an end-to-end DNN architecture named online three-dimensional path planning network (OTDPP-Net) is designed to learn 3D local path planning policies. It determines actions in 3D space based on multiple value iteration computations approximated by recurrent 2D convolutional neural networks. Moreover, a path planning framework is also developed to realize near-optimal real-time online path planning. The effectiveness of the proposed planner is further improved by a switching scheme, and the path quality is optimized by line-of-sight checks. Both virtual and real-world experimental results demonstrate the remarkable performance of the proposed DNN-based path planner in terms of efficiency, success rate and path quality. Different from existing methods, the computational time and effectiveness of the developed DNN-based path planner are both independent of environmental conditions, which reveals its superiority in large-scale complex environments. A video of our experiments can be found at:https://youtu.be/gb4nSG4hd6s.

Domain adaptation is critical for learning transferable features that effectively reduce the distribution difference among domains. In the era of big data, the availability of large-scale labeled datasets motivates partial domain adaptation (PDA) which deals with adaptation from large source domains to small target domains with less number of classes. In the PDA setting, it is crucial to transfer relevant source samples and eliminate irrelevant ones to mitigate negative transfer. In this paper, we propose a deep reinforcement learning based source data selector for PDA, which is capable of eliminating less relevant source samples automatically to boost existing adaptation methods. It determines to either keep or discard the source instances based on their feature representations so that more effective knowledge transfer across domains can be achieved via filtering out irrelevant samples. As a general module, the proposed DRL-based data selector can be integrated into any existing domain adaptation or partial domain adaptation models. Extensive experiments on several benchmark datasets demonstrate the superiority of the proposed DRL-based data selector which leads to state-of-the-art performance for various PDA tasks.

Segmenting the movement path is an essential requirement of intelligent mobile robots. It assists intelligent systems in gaining a better understanding of the scene and identifying re-visited spaces. Moreover, it helps intelligent robots quantize the wide scene into sub-spaces that visually represent the same content-for instance, distinguishing rooms and kitchen in an indoor environment. This paper proposes an unsupervised approach to understand the transition of the scene while a robot is moving and helps to extract sub-spaces that visually represent a similar environment. The proposed approach benefits from a pre-trained deep network architecture to extract a description (feature representation) for the mobile robot's visual information at each time step. Then, based on the pairwise distance of the feature representations, sub-spaces of the scene and transition points are extracted.

The most significant part of any autonomous intelligent robot is the localization module that gives the robot knowledge about its position and orientation. This knowledge assists the robot to move to the location of its desired goal and complete its task. Visual Odometry (VO) measures the displacement of the robots’ camera in consecutive frames which results in the estimation of the robot position and orientation. Deep Learning, nowadays, helps to learn rich and informative features for the problem of VO to estimate frame-by-frame camera movement. Recent Deep Learning-based VO methods train an end-by-end network to solve VO as a regression problem directly without visualizing and sensing the label of training data in the training procedure. In this paper, a new approach to train Convolutional Neural Networks (CNNs) for the regression problems, such as VO, is proposed. The proposed method first changes the problem to a classification problem to learn different subspaces with similar observations. After solving the classification problem, the problem converts to the original regression problem to solve using the knowledge achieved by solving the classification problem. This approach helps CNN to solve regression problem globally in a local domain learned in the classification step, and improves the performance of the regression module for approximately 10%.

Inertial measurement units (IMUs) suffer from bias and measurement noise, which makes it much more complicated to tackle the problem of inertial odometry (IO). Due to the error propagation over time, while estimating robot position, an inaccurate estimation or a small error will cause the odometry and a localization system unreliable and unusable in a split of seconds. This paper presents a novel triple-channel deep IO network architecture based on the physical and mathematical models of IMUs. The proposed method simulates the noise model in the training phase and becomes robust to noise during testing. Besides, the proposed network architecture also considers the time interval between two consecutive IMU readings (sampling time) so that it is robust to the change of IMU frequency and the missing of IMU information. To the best of our knowledge, this paper is the first work reviewing and analyzing the existing IO methods used by the deep-learning-based visual-IO approaches. The proposed network architecture outperforms all the existing solutions on the IMU readings of the challenging Micro Aerial Vehicle dataset and improves the accuracy by approximately 25%.