OCMDP: Observation-Constrained Markov Decision Process

In many practical applications, decision-making processes must balance the costs of acquiring information with the benefits it provides. Traditional control systems often assume full observability, an unrealistic assumption when observations are expensive. We tackle the challenge of simultaneously learning observation and control strategies in such cost-sensitive environments by introducing the Observation-Constrained Markov Decision Process (OCMDP), where the policy influences the observability of the true state. To manage the complexity arising from the combined observation and control actions, we develop an iterative, model-free deep reinforcement learning algorithm that separates the sensing and control components of the policy. This decomposition enables efficient learning in the expanded action space by focusing on when and what to observe, as well as determining optimal control actions, without requiring knowledge of the environment's dynamics. We validate our approach on a simulated diagnostic task and a realistic healthcare environment using HeartPole. Given both scenarios, the experimental results demonstrate that our model achieves a substantial reduction in observation costs on average, significantly outperforming baseline methods by a notable margin in efficiency.

Detect an Object At Once without Fine-tuning

When presented with one or a few photos of a previously unseen object, humans can instantly recognize it in different scenes. Although the human brain mechanism behind this phenomenon is still not fully understood, this work introduces a novel technical realization of this task. It consists of two phases: (1) generating a Similarity Density Map (SDM) by convolving the scene image with the given object image patch(es) so that the highlight areas in the SDM indicate the possible locations; (2) obtaining the object occupied areas in the scene through a Region Alignment Network (RAN). The RAN is constructed on a backbone of Deep Siamese Network (DSN), and different from the traditional DSNs, it aims to obtain the object accurate regions by regressing the location and area differences between the ground truths and the predicted ones indicated by the highlight areas in SDM. By pre-learning from labels annotated in traditional datasets, the SDM-RAN can detect previously unknown objects without fine-tuning. Experiments were conducted on the MS COCO, PASCAL VOC datasets. The results indicate that the proposed method outperforms state-of-the-art methods on the same task.

DistRL: An Asynchronous Distributed Reinforcement Learning Framework for On-Device Control Agents

On-device control agents, especially on mobile devices, are responsible for operating mobile devices to fulfill users' requests, enabling seamless and intuitive interactions. Integrating Multimodal Large Language Models (MLLMs) into these agents enhances their ability to understand and execute complex commands, thereby improving user experience. However, fine-tuning MLLMs for on-device control presents significant challenges due to limited data availability and inefficient online training processes. This paper introduces DistRL, a novel framework designed to enhance the efficiency of online RL fine-tuning for mobile device control agents. DistRL employs centralized training and decentralized data acquisition to ensure efficient fine-tuning in the context of dynamic online interactions. Additionally, the framework is backed by our tailor-made RL algorithm, which effectively balances exploration with the prioritized utilization of collected data to ensure stable and robust training. Our experiments show that, on average, DistRL delivers a 3X improvement in training efficiency and enables training data collection 2.4X faster than the leading synchronous multi-machine methods. Notably, after training, DistRL achieves a 20% relative improvement in success rate compared to state-of-the-art methods on general Android tasks from an open benchmark, significantly outperforming existing approaches while maintaining the same training time. These results validate DistRL as a scalable and efficient solution, offering substantial improvements in both training efficiency and agent performance for real-world, in-the-wild device control tasks.

Neural Surface Reconstruction and Rendering for LiDAR-Visual Systems

This paper presents a unified surface reconstruction and rendering framework for LiDAR-visual systems, integrating Neural Radiance Fields (NeRF) and Neural Distance Fields (NDF) to recover both appearance and structural information from posed images and point clouds. We address the structural visible gap between NeRF and NDF by utilizing a visible-aware occupancy map to classify space into the free, occupied, visible unknown, and background regions. This classification facilitates the recovery of a complete appearance and structure of the scene. We unify the training of the NDF and NeRF using a spatial-varying scale SDF-to-density transformation for levels of detail for both structure and appearance. The proposed method leverages the learned NDF for structure-aware NeRF training by an adaptive sphere tracing sampling strategy for accurate structure rendering. In return, NeRF further refines structural in recovering missing or fuzzy structures in the NDF. Extensive experiments demonstrate the superior quality and versatility of the proposed method across various scenarios. To benefit the community, the codes will be released at \url{https://github.com/hku-mars/M2Mapping}.

Towards Real-time Scalable Dense Mapping using Robot-centric Implicit Representation

Real-time and high-quailty dense mapping is essential for robots to perform fine tasks. However, most existing methods can not achieve both speed and quality. Recent works have shown that implicit neural representations of 3D scenes can produce remarkable results, but they are limited to small scenes and lack real-time performance. To address these limitations, we propose a real-time scalable mapping method using robot-centric implicit representation. We train implicit features with a multi-resolution local map and decode them as signed distance values through a shallow neural network. We maintain the learned features in a scalable manner using a global map that consists of a hash table and a submap set. We exploit the characteristics of the local map to achieve highly efficient training and mitigate the catastrophic forgetting problem in incremental implicit mapping. Extensive experiments validate that our method outperforms existing methods in reconstruction quality, real-time performance, and applicability. The code of our system will be available at \url{https://github.com/HITSZ-NRSL/RIM.git}.

RGB-D Inertial Odometry for a Resource-Restricted Robot in Dynamic Environments

Current simultaneous localization and mapping (SLAM) algorithms perform well in static environments but easily fail in dynamic environments. Recent works introduce deep learning-based semantic information to SLAM systems to reduce the influence of dynamic objects. However, it is still challenging to apply a robust localization in dynamic environments for resource-restricted robots. This paper proposes a real-time RGB-D inertial odometry system for resource-restricted robots in dynamic environments named Dynamic-VINS. Three main threads run in parallel: object detection, feature tracking, and state optimization. The proposed Dynamic-VINS combines object detection and depth information for dynamic feature recognition and achieves performance comparable to semantic segmentation. Dynamic-VINS adopts grid-based feature detection and proposes a fast and efficient method to extract high-quality FAST feature points. IMU is applied to predict motion for feature tracking and moving consistency check. The proposed method is evaluated on both public datasets and real-world applications and shows competitive localization accuracy and robustness in dynamic environments. Yet, to the best of our knowledge, it is the best-performance real-time RGB-D inertial odometry for resource-restricted platforms in dynamic environments for now. The proposed system is open source at: https://github.com/HITSZ-NRSL/Dynamic-VINS.git

Active Implicit Object Reconstruction using Uncertainty-guided Next-Best-View Optimziation

Actively planning sensor views during object reconstruction is essential to autonomous mobile robots. This task is usually performed by evaluating information gain from an explicit uncertainty map. Existing algorithms compare options among a set of preset candidate views and select the next-best-view from them. In contrast to these, we take the emerging implicit representation as the object model and seamlessly combine it with the active reconstruction task. To fully integrate observation information into the model, we propose a supervision method specifically for object-level reconstruction that considers both valid and free space. Additionally, to directly evaluate view information from the implicit object model, we introduce a sample-based uncertainty evaluation method. It samples points on rays directly from the object model and uses variations of implicit function inferences as the uncertainty metrics, with no need for voxel traversal or an additional information map. Leveraging the differentiability of our metrics, it is possible to optimize the next-best-view by maximizing the uncertainty continuously. This does away with the traditionally-used candidate views setting, which may provide sub-optimal results. Experiments in simulations and real-world scenes show that our method effectively improves the reconstruction accuracy and the view-planning efficiency of active reconstruction tasks. The proposed system is going to open source at https://github.com/HITSZ-NRSL/ActiveImplicitRecon.git.