Generalization in visual reinforcement learning (VRL) aims to learn agents that can adapt to test environments with unseen visual distractions. Despite advances in robust representations learning, many methods do not take into account the essential downstream task of sequential decision-making. This leads to representations that lack critical long-term information, impairing decision-making abilities in test environments. To tackle this problem, we propose a novel robust action-value representation learning (ROUSER) under the information bottleneck (IB) framework. ROUSER learns robust representations to capture long-term information from the decision-making objective (i.e., action values). Specifically, ROUSER uses IB to encode robust representations by maximizing their mutual information with action values for long-term information, while minimizing mutual information with state-action pairs to discard irrelevant features. As action values are unknown, ROUSER proposes to decompose robust representations of state-action pairs into one-step rewards and robust representations of subsequent pairs. Thus, it can use known rewards to compute the loss for robust representation learning. Moreover, we show that ROUSER accurately estimates action values using learned robust representations, making it applicable to various VRL algorithms. Experiments demonstrate that ROUSER outperforms several state-of-the-art methods in eleven out of twelve tasks, across both unseen background and color distractions.

Real-world offline datasets are often subject to data corruptions (such as noise or adversarial attacks) due to sensor failures or malicious attacks. Despite advances in robust offline reinforcement learning (RL), existing methods struggle to learn robust agents under high uncertainty caused by the diverse corrupted data (i.e., corrupted states, actions, rewards, and dynamics), leading to performance degradation in clean environments. To tackle this problem, we propose a novel robust variational Bayesian inference for offline RL (TRACER). It introduces Bayesian inference for the first time to capture the uncertainty via offline data for robustness against all types of data corruptions. Specifically, TRACER first models all corruptions as the uncertainty in the action-value function. Then, to capture such uncertainty, it uses all offline data as the observations to approximate the posterior distribution of the action-value function under a Bayesian inference framework. An appealing feature of TRACER is that it can distinguish corrupted data from clean data using an entropy-based uncertainty measure, since corrupted data often induces higher uncertainty and entropy. Based on the aforementioned measure, TRACER can regulate the loss associated with corrupted data to reduce its influence, thereby enhancing robustness and performance in clean environments. Experiments demonstrate that TRACER significantly outperforms several state-of-the-art approaches across both individual and simultaneous data corruptions.

Adding building labels greatly improves the recognizability of buildings and the readability of three-dimensional (3D) city scenes. However, building label placement is much more complex in 3D scenes than in two-dimensional (2D) maps. The annotation effect is influenced by the attributes of the 3D label, building visibility, and the spatial relationship between the building and viewpoint. In this context, automatically generating building labels for 3D scenes during interactions requires highly complex computations. By contrast, evaluating candidate labels and then selecting the suitable label for each building can be effectively implemented. This paper introduces an approach for labeling buildings in 3D scenes based on evaluations of label candidates. The proposed method predefines a candidate label set for each building. These candidates are then evaluated in terms of their attributes and the relationship between the labels and viewpoint at runtime. The best candidate label, or a situational alternative for each building, is then placed in order of comprehensive label priority to avoid annotation conflicts. A series of experiments demonstrate that this method effectively enhances the correlation of labels and buildings, improves interactive efficiency, and realizes a viable global label layout.

High error rate is critical to the wireless video transmission. Our paper tackles the problem of robust video streaming over error-prone channels. Basing on an autoresilient multiple-description coding method and a multi-path transmission strategy, further study is done on the computational complexity of rate-distortion optimization for video coding and transmission. A unitary-optimized operation of the whole system is then proposed after the R-D calculation. Experiment results show that video streaming with our unitary-optimized operation gains better playback quality especially when transmitted through error-prone mobile channel.