Although Gaussian scene representation has achieved remarkable success in tracking and mapping, most existing methods are confined to single-agent systems. Current multi-agent solutions typically rely on centralized architectures, which struggle to account for communication bandwidth constraints. Furthermore, the inherent depth ambiguity of 3D Gaussian splatting poses notable challenges in maintaining geometric consistency. To address these challenges, we introduce CoMA-SLAM, the first distributed multi-agent Gaussian SLAM framework. By leveraging 2D Gaussian surfels and robust initialization strategy, CoMA-SLAM enhances tracking accuracy and geometry consistency. It efficiently manages communication bandwidth while dynamically scaling with the number of agents. Through the integration of intra- and inter-loop closure, distributed keyframe optimization and submap centric update, our framework ensures global consistency and robustly alignment. Synthetic and real-world experiments demonstrate that CoMA-SLAM outperforms state-of-the-art methods in pose accuracy, rendering fidelity, and geometric consistency while maintaining competitive efficiency across distributed multi-agent systems. Notably, by avoiding data transmission to a centralized server, our method reduces communication bandwidth by 99.8% compared to centralized approaches.

Jiaqi Li, Guangming Wang, Shuntian Zheng, Minzhe Ni, Xiaoman Lu, Guanghui Ye, Yu Guan. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

3D scene flow represents the dense per-point motion field in dynamic scenes, playing a crucial role in various downstream tasks, including motion segmentation, dynamic scene reconstruction, 4D content generation, etc. However, previous regression-based works commonly suffer from unreliable correlations caused by locally constrained search ranges and struggle with the absence of timely feedback regarding the flow estimation uncertainty during training. To address these challenges, we propose a novel uncertainty-aware network for scene flow estimation, termed DifFlow3D, based on the conditional probabilistic diffusion model. Hierarchical diffusion-based flow estimation blocks are designed to enhance the correlation robustness and resilience to challenging cases, e.g., dynamics, noisy inputs, repetitive patterns, etc. To mitigate the generation diversity, three key flow-related features are leveraged as conditions in our diffusion model. Furthermore, we develop an uncertainty estimation module within diffusion to assess the reliability of estimated scene flow dynamically. A Hidden State Denoising strategy (HSD) is also introduced to further boost the stability of the reverse denoising process. Extensive experiments conducted on four scene flow datasets, including both synthetic and real-world datasets (FlyingThings3D, KITTI 2015, Argoverse, and Waymo Open), demonstrate the superiority of our proposed DifFlow3D. Compared to prior state-of-the-art methods, DifFlow3D has 26.0%, 36.4%, 35.3%, and 17.7% EPE3D reduction respectively across four datasets. Only trained on the synthetic FlyingThings3D dataset, our method achieves an unprecedented millimeter-level accuracy (0.0070 m EPE3D) on the real-scene KITTI dataset, highlighting its exceptional generalization capability. Additionally, our diffusion-based refinement paradigm can be seamlessly integrated as a plug-and-play module into existing scene flow networks, significantly enhancing their estimation accuracy. We also introduce our pre-trained scene flow estimator as explicit motion priors into the novel dynamic LiDAR view synthesis task, which validates its great potential for improving the 4D LiDAR reconstruction performance.

We propose SNI-SLAM++, a tightly-coupled semantic SLAM system utilizing neural implicit representation, that simultaneously performs accurate semantic mapping, high-quality surface reconstruction, and robust camera tracking. Our system tightly integrates visual appearance, geometry, and semantics through five key components: (i) We introduce hierarchical semantic representation to allow multi-level semantic comprehension for top-down structured semantic mapping of the scene. (ii) To fully utilize the correlation between multiple attributes of the environment, we integrate appearance, geometry and semantic features through cross-attention for feature collaboration. This strategy enables a more multifaceted understanding of the environment, thereby allowing SNI-SLAM++ to remain robust even when single attribute is defective. (iii) We design an internal fusion-based decoder to obtain semantic, RGB, and Truncated Signed Distance Field (TSDF) values from multi-level features for accurate decoding. (iv) We introduce a semantics-coupled tracking framework that tightly incorporates semantic constraints for camera pose estimation in neural implicit SLAM. This framework leverages the multi-view consistency of semantics to construct a pose graph and perform semantic loop closure optimization, enabling robust tracking. (v) We propose a feature loss to update the scene representation at the feature level. Compared with low-level losses such as RGB loss and depth loss, our feature loss is capable of guiding the network optimization on a higher level. Our SNI-SLAM++ demonstrates superior performance over all recent visual SLAM methods in terms of mapping and tracking accuracy on the datasets of Replica, ScanNet, TUM-RGBD, and ScanNet++, while also showing excellent capabilities in accurate semantic segmentation and 3D semantic mapping.

With the rapid development of multimodal data fusion based on deep learning, semantic segmentation using the complementary information from the camera and LiDAR sensor can enhance the environmental perception of autonomous driving. However, most fusion methods overlook critical issues arising during the fusion process, including: first, insufficient fusion effectiveness in one-to-one scenarios due to a lack of precise feature alignment; second, low efficiency in the fusion process and a large computational burden on the model. To address these issues, we propose a trans-modal fusion method based on state space model and deformable transformer module. The proposed trans-modal fusion method makes use of the 2-D prior information of the image, such as texture and appearance information. By distilling knowledge from fused 2D–3D representations, it significantly enhances the feature extraction capability of the 3-D point cloud. First, we design a lightweight global gap adjustment attention module based on a linear state-space model, named global gap attention, to better bridge the information gap between point clouds and images. Then, to enhance the efficiency of trans-modal fusion, we adopt a local cross fusion module, which utilizes a many-to-many fusion approach to achieve more accurate feature alignment. Finally, to strengthen the channel information of point cloud features, we introduce an efficient channel attention during the feature extraction process. The experimental results show that the proposed method achieves a better segmentation accuracy compared with baseline. We show superiority of our method on two popular urban datasets SemanticKITTI and nuScenes (achieving mean Intersection over Union (mIoU) scores of 70.9% and 80.4%, respectively).

Sim-to-Real refers to the process of transferring policies learned in simulation to the real world, which is crucial for achieving practical robotics applications. However, recent Sim2real methods either rely on a large amount of augmented data or large learning models, which is inefficient for specific tasks. In recent years, with the emergence of radiance field reconstruction methods, especially 3D Gaussian splatting, it has become possible to construct realistic real-world scenes. To this end, we propose RL-GSBridge, a novel real-to-sim-to-real framework which incorporates 3D Gaussian Splatting into the conventional RL simulation pipeline, enabling zero-shot sim-to-real transfer for vision-based deep reinforcement learning. We introduce a mesh-based 3D GS method with soft binding constraints, enhancing the rendering quality of mesh models. Then utilizing a GS editing approach to synchronize the rendering with the physics simulator, RL-GSBridge could reflect the visual interactions of the physical robot accurately. Through a series of sim-to-real experiments, including grasping and pick-and-place tasks, we demonstrate that RL-GSBridge maintains a satisfactory success rate in real-world task completion during sim-to-real transfer. Furthermore, a series of rendering metrics and visualization results indicate that our proposed mesh-based 3D GS reduces artifacts in unstructured objects, demonstrating more realistic rendering performance.

Recent research on Simultaneous Localization and Mapping (SLAM) based on implicit representation has shown promising results in indoor environments. However, some challenges remain: the limited scene representation capability of implicit encoding, the uncertainty in the rendering process from implicit representations, and the disruption of consistency by dynamic objects. To address these challenges, we propose a dynamic visual SLAM system based on local-global fusion neural implicit representation, named DVN-SLAM. To improve the scene representation capability, we introduce a local-global fusion neural implicit representation that enables the construction of an implicit map while considering both global structure and local details. To tackle uncertainties arising from the rendering process, we design an information concentration loss for optimization, aiming to concentrate scene information on object surfaces. The proposed DVN-SLAM achieves competitive performance in localization and mapping across multiple datasets. More importantly, DVN-SLAM demonstrates robustness without semantic and optical flow prior in dynamic scenes, which sets it apart from other NeRF-based methods.

This paper presents a novel collaborative online dense mapping system for multiple Unmanned Aerial Vehicles (UAVs). The system confers two primary benefits: it facilitates simultaneous UAVs co-localization and real-time dense map reconstruction, and it recovers the metric scale even in GNSS-denied conditions. To achieve these advantages, Ultrawideband (UWB) measurements, monocular Visual Odometry (VO), and co-visibility observations are jointly employed to recover both relative positions and global UAV poses, thereby ensuring optimality at both local and global scales. In the proposed methodology, a two-stage optimization strategy is proposed to reduce optimization burden. Initially, relative Sim3 transformations among UAVs are swiftly estimated, with UWB measurements facilitating metric scale recovery in the absence of GNSS. Subsequently, a global pose optimization is performed to effectively mitigate cumulative drift. By integrating UWB, VO, and co-visibility data within this framework, both local geometric consistency and global pose accuracy are robustly maintained. Through comprehensive simulation and empirical real-world testing, we demonstrate that our system not only improves UAV positioning accuracy in challenging scenarios but also facilitates the high-quality, online integration of dense point clouds in large-scale areas. This research offers valuable contributions and practical techniques for precise, real-time map reconstruction using an autonomous UAV fleet, particularly in GNSS-denied environments.

Moving object segmentation plays a vital role in understanding dynamic visual environments. While existing methods rely on multi-frame image sequences to identify moving objects, single-image MOS is critical for applications like motion intention prediction and handling camera frame drops. However, segmenting moving objects from a single image remains challenging for existing methods due to the absence of temporal cues. To address this gap, we propose MovSAM, the first framework for single-image moving object segmentation. MovSAM leverages a Multimodal Large Language Model (MLLM) enhanced with Chain-of-Thought (CoT) prompting to search the moving object and generate text prompts based on deep thinking for segmentation. These prompts are cross-fused with visual features from the Segment Anything Model (SAM) and a Vision-Language Model (VLM), enabling logic-driven moving object segmentation. The segmentation results then undergo a deep thinking refinement loop, allowing MovSAM to iteratively improve its understanding of the scene context and inter-object relationships with logical reasoning. This innovative approach enables MovSAM to segment moving objects in single images by considering scene understanding. We implement MovSAM in the real world to validate its practical application and effectiveness for autonomous driving scenarios where the multi-frame methods fail. Furthermore, despite the inherent advantage of multi-frame methods in utilizing temporal information, MovSAM achieves state-of-the-art performance across public MOS benchmarks, reaching 92.5% on J&F. Our implementation will be available at https://github.com/IRMVLab/MovSAM.

We propose SemGauss-SLAM, a dense semantic SLAM system utilizing 3D Gaussian representation, that enables accurate 3D semantic mapping, robust camera tracking, and high-quality rendering simultaneously. In this system, we incorporate semantic feature embedding into 3D Gaussian representation, which effectively encodes semantic information within the spatial layout of the environment for precise semantic scene representation. Furthermore, we propose feature-level loss for updating 3D Gaussian representation, enabling higher-level guidance for 3D Gaussian optimization. In addition, to reduce cumulative drift in tracking and improve semantic reconstruction accuracy, we introduce semantic-informed bundle adjustment. By leveraging multi-frame semantic associations, this strategy enables joint optimization of 3D Gaussian representation and camera poses, resulting in low-drift tracking and accurate semantic mapping. Our SemGauss-SLAM demonstrates superior performance over existing radiance field-based SLAM methods in terms of mapping and tracking accuracy on Replica and ScanNet datasets, while also showing excellent capabilities in high-precision semantic segmentation and dense semantic mapping. Code will be available at https://github.com/IRMVLab/SemGauss-SLAM.

In this paper, we introduce G2-Mapping, a novel method to comprehensively support online monocular, RGB-D, and LiDAR-Inertial-Visual systems, employing 3D gaussian points as scene representation. There are several issues when applying 3d gaussian splatting (3DGS) techniques to simultaneous localization and mapping (SLAM) 1) for monocular, the lack of depth information makes scene initialization difficult and large baseline positioning challenging; 2) differentiable rendering with respect to depth and pose has not been implemented in 3DGS, making it difficult to directly apply to the SLAM system; 3) strategy for updating the scene with incoming online frames is not present, which may lead to memory overflow. In order to overcome problems mentioned above, we formulate a mathematical derivation and propose a differentiable rendering approach that leverages both depth and color to optimize the scene and pose. We introduce a simplified odometry that provides a metric depth estimation for monocular and enhance the low-overlap scene availability. A scale consistency and uncertainty weighted optimization is further proposed to eliminates the impact of inaccurate depth prediction. Our proposed scene updating strategy effectively prevents rapid memory growth. Tracking and mapping are performed alternatively to achieve precise localization and synchronous high-fidelity map reconstruction. Extensive experiments demonstrate that our G2-Mapping surpasses feature-based SLAM in localization precision and exceeds state-of-the-art neural SLAM methods in the fidelity of view synthesis. Note to Practitioners—This paper is dedicated to tackling the efficiency challenges in multi-source SLAM and map reconstruction, aiming to generate pose and high-fidelity maps synchronously. We introduce G2-Mapping, a novel framework that leverages the power of 3D Gaussian points for scene representation, offering a universal solution for monocular, RGB-D, and LiDAR-Inertial-Visual systems. By developing a comprehensive differentiable renderer and presenting a strategy for dynamic scene updating, G2-Mapping significantly advances the state-of-the-art in localization precision and view synthesis fidelity. Although the approach is highly promising, it currently relies on the accuracy of depth prediction networks and requires further optimization for handling sparse LiDAR data. Future research will focus on enhancing these aspects, aiming to seamlessly integrate G2-Mapping into practical applications within robotics, autonomous vehicles, and augmented reality, where robust and efficient SLAM solutions are paramount.

Vision-based occupancy prediction, also known as 3D Semantic Scene Completion (SSC), presents a significant challenge in computer vision. Previous methods, confined to onboard processing, struggle with simultaneous geometric and semantic estimation, continuity across varying viewpoints, and single-view occlusion. Our paper introduces OccFiner, a novel offboard framework designed to enhance the accuracy of vision-based occupancy predictions. OccFiner operates in two hybrid phases: 1) a multi-to-multi local propagation network that implicitly aligns and processes multiple local frames for correcting onboard model errors and consistently enhancing occupancy accuracy across all distances. 2) the region-centric global propagation, focuses on refining labels using explicit multi-view geometry and integrating sensor bias, particularly for increasing the accuracy of distant occupied voxels. Extensive experiments demonstrate that OccFiner improves both geometric and semantic accuracy across various types of coarse occupancy, setting a new state-of-the-art performance on the SemanticKITTI dataset. Notably, OccFiner significantly boosts the performance of vision-based SSC models, achieving accuracy levels competitive with established LiDAR-based onboard SSC methods. Furthermore, OccFiner is the first to achieve automatic annotation of SSC in a purely vision-based approach. Quantitative experiments prove that OccFiner successfully facilitates occupancy data loop-closure in autonomous driving. Additionally, we quantitatively and qualitatively validate the superiority of the offboard approach on city-level SSC static maps. The source code will be made publicly available at https://github.com/MasterHow/OccFiner

3D scene flow represents the 3D motion of each point in the point cloud, which is a base 3D perception task for autonomous driving, like optical flow for 2D images. As non-learning methods are often inefficient or struggled to learn accurate correspondence in complex 3D real world, recent works turn to supervised learning methods, which require ground truth labels. However, acquiring the ground truth of 3D scene flow is challenging mainly due to the lack of sensors capable of capturing point-level motion and the complexity of accurately tracking each point in real-world environments. Therefore, it is important to resort to self-supervised methods, which do not require ground truth labels. In this paper, a novel unsupervised learning method of scene flow with LiDAR odometry is proposed, which enables the scene flow network can be trained directly on real-world LiDAR data without scene flow labels. In this structure, supervised odometry provides a more accurate shared cost volume for the interframe association of 3D scene flow. In addition, because static and occluded points are more suitable for using the pose transform while dynamic and non-occluded points are more suitable for using the scene flow transform, a static mask and an occlusion mask are designed to classify the states of points and a mask-weighted warp layer is proposed to transform source points in a divide-and-conquer manner. The experiments demonstrate that the divide-and-conquer strategy makes the predicted scene flow more accurate. The experiment results compared to other methods also show the application ability of our proposed method to real-world data. Our source codes are released at:https://github.com/IRMVLab/PSFNet.

Robot localization using a built map is essential for a variety of tasks including accurate navigation and mobile manipulation. A popular approach to robot localization is based on image-to-point cloud registration, which combines illumination-invariant LiDAR-based mapping with economical image-based localization. However, the recent works for image-to-point cloud registration either divide the registration into separate modules or project the point cloud to the depth image to register the RGB and depth images. In this paper, we present I2PNet, a novel end-to-end 2D-3D registration network, which directly registers the raw 3D point cloud with the 2D RGB image using differential modules with a united target. The 2D-3D cost volume module for differential 2D-3D association is proposed to bridge feature extraction and pose regression. The soft point-to-pixel correspondence is implicitly constructed on the intrinsic-independent normalized plane in the 2D-3D cost volume module. Moreover, we introduce an outlier mask prediction module to filter the outliers in the 2D-3D association before pose regression. Furthermore, we propose the coarse-to-fine 2D-3D registration architecture to increase localization accuracy. Extensive localization experiments are conducted on the KITTI, nuScenes, M2DGR, Argoverse, Waymo, and Lyft5 datasets. The results demonstrate that I2PNet outperforms the state-of-the-art by a large margin and has a higher efficiency than the previous works. Moreover, we extend the application of I2PNet to the camera-LiDAR online calibration and demonstrate that I2PNet outperforms recent approaches on the online calibration task. Source codes are released athttps://github.com/IRMVLab/I2PNet.

Learning 3D scene flow from LiDAR point clouds presents significant difficulties, including poor generalization from synthetic datasets to real scenes, scarcity of real-world 3D labels, and poor performance on real sparse Li-DAR point clouds. We present a novel approach from the perspective of auto-labelling, aiming to generate a large number of 3D scene flow pseudo labels for real-world Li-DAR point clouds. Specifically, we employ the assumption of rigid body motion to simulate potential object-level rigid movements in autonomous driving scenarios. By updating different motion attributes for multiple anchor boxes, the rigid motion decomposition is obtained for the whole scene. Furthermore, we developed a novel 3D scene flow data augmentation method for global and local motion. By perfectly synthesizing target point clouds based on augmented motion parameters, we easily obtain lots of 3D scene flow labels in point clouds highly consistent with real scenarios. On multiple real-world datasets including LiDAR KITTI, nuScenes, and Argoverse, our method outperforms all previous supervised and unsupervised methods without requiring manual labelling. Impressively, our method achieves a tenfold reduction in EPE3D metric on the LiDAR KITTI dataset, reducing it from 0.190m to a mere 0.008m error.

Scene flow estimation, which aims to predict per-point 3D displacements of dynamic scenes, is a fundamen-tal task in the computer vision field. However, previ-ous works commonly suffer from unreliable correlation caused by locally constrained searching ranges, and struggle with accumulated inaccuracy arising from the coarse-to-fine structure. To alleviate these problems, we propose a novel uncertainty-aware scene flow estimation network(DifFlow3D) with the diffusion probabilistic model. Iter-ative diffusion-based refinement is designed to enhance the correlation robustness and resilience to challenging cases, e.g. dynamics, noisy inputs, repetitive patterns, etc. To re-strain the generation diversity, three key flow-related features are leveraged as conditions in our diffusion model. Furthermore, we also develop an uncertainty estimation module within diffusion to evaluate the reliability of esti-mated scene flow. Our DifFlow3D achieves state-of-the-art performance, with 24.0% and 29.1% EPE3D reduction respectively on FlyingThings3D and KITTI 2015 datasets. Notably, our method achieves an unprecedented millimeter-level accuracy (O.0078m in EPE3D) on the KITTI dataset. Additionally, our diffusion-based refinement paradigm can be readily integrated as a plug-and-play module into ex-isting scene flow networks, significantly increasing their estimation accuracy. Codes are released at https:// github.com/IRMVLab/DifFlow3D.

We propose SNI-SLAM, a semantic SLAM system utilizing neural implicit representation, that simultaneously performs accurate semantic mapping, high-quality surface reconstruction, and robust camera tracking. In this system, we introduce hierarchical semantic representation to allow multi-level semantic comprehension for top-down structured semantic mapping of the scene. In addition, to fully utilize the correlation between multiple attributes of the environment, we integrate appearance, geometry and semantic features through cross-attention for feature collaboration. This strategy enables a more multifaceted understanding of the environment, thereby allowing SNI-SLAM to remain robust even when single attribute is defective. Then, we design an internal fusion-based decoder to obtain semantic, RGB, Truncated Signed Distance Field (TSDF) values from multi-level features for accurate decoding. Furthermore, we propose a feature loss to update the scene representation at the feature level. Compared with low-level losses such as RGB loss and depth loss, our feature loss is capable of guiding the network optimization on a higher-level. Our SNI-SLAM method demonstrates superior performance over all recent NeRF-based SLAM methods in terms of mapping and tracking accuracy on Replica and ScanNet datasets, while also showing excellent capabilities in accurate semantic segmentation and real-time semantic mapping. Codes will be available at https://github.com/IRMVLab/SNI-SLAM.

Localization using a monocular camera in the pre-built LiDAR point cloud map has drawn increasing attention in the field of autonomous driving and mobile robotics. However, there are still many challenges (e.g. difficulties of map storage, poor localization robustness in large scenes) in accurately and efficiently implementing cross-modal localization. To solve these problems, a novel pipeline termed LHMap-loc is proposed, which achieves accurate and efficient monocular localization in LiDAR maps. Firstly, feature encoding is carried out on the original LiDAR point cloud map by generating offline heat point clouds, by which the size of the original LiDAR map is compressed. Then, an end-to-end online pose regression network is designed based on optical flow estimation and spatial attention to achieve real-time monocular visual localization in a pre-built map. In addition, a series of experiments have been conducted to prove the effectiveness of the proposed method. Our code is available at: https://github.com/IRMVLab/LHMap-loc.

This paper introduces a 3D point cloud sequence learning model based on inconsistent spatio-temporal propagation for LiDAR odometry, termed DSLO. It consists of a pyramid structure with a spatial information reuse strategy, a sequential pose initialization module, a gated hierarchical pose refinement module, and a temporal feature propagation module. First, spatial features are encoded using a point feature pyramid, with features reused in successive pose estimations to reduce computational overhead. Second, a sequential pose initialization method is introduced, leveraging the high-frequency sampling characteristic of LiDAR to initialize the LiDAR pose. Then, a gated hierarchical pose refinement mechanism refines poses from coarse to fine by selectively retaining or discarding motion information from different layers based on gate estimations. Finally, temporal feature propagation is proposed to incorporate the historical motion information from point cloud sequences, and address the spatial inconsistency issue when transmitting motion information embedded in point clouds between frames. Experimental results on the KITTI odometry dataset and Argoverse dataset demonstrate that DSLO outperforms state-of-the-art methods, achieving at least a 15.67% improvement on RTE and a 12.64% improvement on RRE, while also achieving a 34.69% reduction in runtime compared to baseline methods. Our implementation will be available at https://github.com/IRMVLab/DSLO.

LiDAR point cloud semantic segmentation enables the robots to obtain fine-grained semantic information of the surrounding environment. Recently, many works project the point cloud onto the 2D image and adopt the 2D Convolutional Neural Networks (CNNs) or vision transformer for LiDAR point cloud semantic segmentation. However, since more than one point can be projected onto the same 2D position but only one point can be preserved, the previous 2D projection-based segmentation methods suffer from inevitable quantized information loss, which results in incomplete geometric structure, especially for small objects. To avoid quantized information loss, in this paper, we propose a novel spherical frustum structure, which preserves all points projected onto the same 2D position. Additionally, a hash-based representation is proposed for memory-efficient spherical frustum storage. Based on the spherical frustum structure, the Spherical Frustum sparse Convolution (SFC) and Frustum Farthest Point Sampling (F2PS) are proposed to convolve and sample the points stored in spherical frustums respectively. Finally, we present the Spherical Frustum sparse Convolution Network (SFCNet) to adopt 2D CNNs for LiDAR point cloud semantic segmentation without quantized information loss. Extensive experiments on the SemanticKITTI and nuScenes datasets demonstrate that our SFCNet outperforms previous 2D projection-based semantic segmentation methods based on conventional spherical projection and shows better performance on small object segmentation by preserving complete geometric structure. Codes will be available at https://github.com/IRMVLab/SFCNet.

Deep reinforcement learning (DRL) enables robots to perform some intelligent tasks end-to-end. However, there are still many challenges for long-horizon sparse-reward robotic manipulator tasks. On the one hand, a sparse-reward setting causes exploration inefficient. On the other hand, exploration using physical robots is of high cost and unsafe. In this article, we propose a method of learning long-horizon sparse-reward tasks utilizing one or more existing traditional controllers named base controllers in this article. Built upon deep deterministic policy gradients (DDPGs), our algorithm incorporates the existing base controllers into stages of exploration, value learning, and policy update. Furthermore, we present a straightforward way of synthesizing different base controllers to integrate their strengths. Through experiments ranging from stacking blocks to cups, it is demonstrated that the learned state-based or image-based policies steadily outperform base controllers. Compared to previous works of learning from demonstrations, our method improves sample efficiency by orders of magnitude and improves performance. Overall, our method bears the potential of leveraging existing industrial robot manipulation systems to build more flexible and intelligent controllers.

Recently, transformer architecture has gained great success in the computer vision community, such as image classification, object detection, etc. Nonetheless, its application for 3D vision remains to be explored, given that point cloud is inherently sparse, irregular, and unordered. Furthermore, existing point transformer frameworks usually feed raw point cloud of N×3 dimension into transformers, which limits the point processing scale because of their quadratic computational costs to the input size N. In this paper, we rethink the structure of point transformer. Instead of directly applying transformer to points, our network (TransLO) can process tens of thousands of points simultaneously by projecting points onto a 2D surface and then feeding them into a local transformer with linear complexity. Specifically, it is mainly composed of two components: Window-based Masked transformer with Self Attention (WMSA) to capture long-range dependencies; Masked Cross-Frame Attention (MCFA) to associate two frames and predict pose estimation. To deal with the sparsity issue of point cloud, we propose a binary mask to remove invalid and dynamic points. To our knowledge, this is the first transformer-based LiDAR odometry network. The experiment results on the KITTI odometry dataset show that our average rotation and translation RMSE achieves 0.500°/100m and 0.993% respectively. The performance of our network surpasses all recent learning-based methods and even outperforms LOAM on most evaluation sequences.Codes will be released on https://github.com/IRMVLab/TransLO.

Although point cloud registration has achieved remarkable advances in object-level and indoor scenes, large-scale registration methods are rarely explored. Challenges mainly arise from the huge point number, complex distribution, and outliers of outdoor LiDAR scans. In addition, most existing registration works generally adopt a two-stage paradigm: They first find correspondences by extracting discriminative local features and then leverage estimators (eg. RANSAC) to filter outliers, which are highly dependent on well-designed descriptors and post-processing choices. To address these problems, we propose an end-to-end transformer network (RegFormer) for large-scale point cloud alignment without any further post-processing. Specifically, a projection-aware hierarchical transformer is proposed to capture long-range dependencies and filter outliers by extracting point features globally. Our transformer has linear complexity, which guarantees high efficiency even for large-scale scenes. Furthermore, to effectively reduce mismatches, a bijective association transformer is designed for regressing the initial transformation. Extensive experiments on KITTI and NuScenes datasets demonstrate that our RegFormer achieves competitive performance in terms of both accuracy and efficiency. Codes are available at https://github.com/IRMVLab/RegFormer.

Robust estimation is a crucial and still challenging task, which involves estimating model parameters in noisy environments. Although conventional sampling consensus-based algorithms sample several times to achieve robustness, these algorithms cannot use data features and historical information effectively. In this paper, we propose RLSAC, a novel Reinforcement Learning enhanced SAmple Consensus framework for end-to-end robust estimation. RLSAC employs a graph neural network to utilize both data and memory features to guide exploring directions for sampling the next minimum set. The feedback of downstream tasks serves as the reward for unsupervised training. Therefore, RL-SAC can avoid differentiating to learn the features and the feedback of downstream tasks for end-to-end robust estimation. In addition, RLSAC integrates a state transition module that encodes both data and memory features. Our experimental results demonstrate that RLSAC can learn from features to gradually explore a better hypothesis. Through analysis, it is apparent that RLSAC can be easily transferred to other sampling consensus-based robust estimation tasks. To the best of our knowledge, RLSAC is also the first method that uses reinforcement learning to sample consensus for end-to-end robust estimation. We release our codes at https://github.com/IRMVLab/RLSAC.

Point clouds are naturally sparse, while image pixels are dense. The inconsistency limits feature fusion from both modalities for point-wise scene flow estimation. Previous methods rarely predict scene flow from the entire point clouds of the scene with one-time inference due to the memory inefficiency and heavy overhead from distance calculation and sorting involved in commonly used farthest point sampling, KNN, and ball query algorithms for local feature aggregation. To mitigate these issues in scene flow learning, we regularize raw points to a dense format by storing 3D coordinates in 2D grids. Unlike the sampling operation commonly used in existing works, the dense 2D representation 1) preserves most points in the given scene, 2) brings in a significant boost of efficiency, and 3) eliminates the density gap between points and pixels, allowing us to perform effective feature fusion. We also present a novel warping projection technique to alleviate the information loss problem resulting from the fact that multiple points could be mapped into one grid during projection when computing cost volume. Sufficient experiments demonstrate the efficiency and effectiveness of our method, outperforming the prior-arts on the FlyingThings3D and KITTI dataset. Our source codes will be released on https://github.com/IRMVLab/DELFlow.

Human drivers are remarkably robust against various unexpected occurring variations and corruptions by understanding temporal changes and traffic scenes. In contrast, the neural network based autonomous navigation system can be easily affected by sensor data anomaly, like occlusion, sensor noise, challenging weather and illumination conditions. Such external disturbances are inevitable in practical driving applications. In this paper, we develop a semi-supervised anomaly detection module to detect the corrupted data while extracting the traffic scenario features. We further introduce an end-to-end robust autonomous navigation framework based on the idea that the consecutive frames of clean data depict a similar traffic scenario and the differences among the sequential data imply the dynamic state changes. By taking into consideration both spatial traffic scenario and temporal environmental variation, the model is able to achieve robust navigation against sensor data corruptions. We conduct experiments in CARLA platform and the evaluation results show the effectiveness of the proposed method.

The main content of the research in this paper is the estimation of depth and pose based on monocular vision and Inertial Measurement Unit (IMU). The usual depth estimation network and pose estimation network require depth ground truth or pose ground truth as a supervised signal for training, while the depth ground truth and pose ground truth are hard to obtain, and monocular vision based depth estimation cannot predict absolute depth. In this paper, with the help of IMU, which is inexpensive and widely used, we can obtain angular velocity and acceleration information. Two new supervision signals are proposed and the calculation expressions are given. Among them, the model trained with acceleration constraint shows a good ability to estimate the absolute depth during the test. It can be considered that the model can estimate the absolute depth. We also derive the method of estimating the scale factor during the test from the acceleration constraint, and also achieve good results as the acceleration constraint does. In addition, this paper also studies the method of using IMU information as pose network input and as selecting conditions. Moreover, it analyzes and discusses the experimental results. At the same time, we also evaluate the effect of the pose estimation of the relevant models. This article starts by reviewing the achievements and deficiencies of the work in this field, combines the use of IMU, puts forward three new methods such as a new loss function, and conducts a test analysis and discussion of relevant indicators on the KITTI data set.

Depth estimation is one of the most important tasks in scene understanding. In the existing joint self-supervised learning approaches of depth-pose estimation, depth estimation and pose estimation networks are independent of each other. They only use the adjacent image frames for pose estimation and lack the use of the estimated geometric information. To enhance the depth-pose association, we propose a monocular multi-frame unsupervised depth estimation framework, named PLPE-Depth. There are a depth estimation network and two pose estimation networks with image input and pseudo-LiDAR input. The main idea of our approach is to use the pseudo-LiDAR reconstructed from the depth map to estimate the pose of adjacent frames. We propose depth re-estimation with a better pose between the image pose and the pseudo-LiDAR pose to improve the accuracy of estimation. Besides, we improve the reconstruction loss and design a pseudo-LiDAR pose enhancement loss to facilitate the joint learning. Our approach enhances the use of the estimated depth information and strengthens the coupling between depth estimation and pose estimation. Experiments on the KITTI dataset show that our depth estimation achieves state-of-the-art performance at low resolution. Our source codes will be released on https://github.com/IRMVLabIPLPE-Depth.

An efficient 3D point cloud learning architecture, named EfficientLO-Net, for LiDAR odometry is first proposed in this article. In this architecture, the projection-aware representation of the 3D point cloud is proposed to organize the raw 3D point cloud into an ordered data form to achieve efficiency. The Pyramid, Warping, and Cost volume (PWC) structure for the LiDAR odometry task is built to estimate and refine the pose in a coarse-to-fine approach. A projection-aware attentive cost volume is built to directly associate two discrete point clouds and obtain embedding motion patterns. Then, a trainable embedding mask is proposed to weigh the local motion patterns to regress the overall pose and filter outlier points. The trainable pose warp-refinement module is iteratively used with embedding mask optimized hierarchically to make the pose estimation more robust for outliers. The entire architecture is holistically optimized end-to-end to achieve adaptive learning of cost volume and mask, and all operations involving point cloud sampling and grouping are accelerated by projection-aware 3D feature learning methods. The superior performance and effectiveness of our LiDAR odometry architecture are demonstrated on KITTI, M2DGR, and Argoverse datasets. Our method outperforms all recent learning-based methods and even the geometry-based approach, LOAM with mapping optimization, on most sequences of KITTI odometry dataset. We open sourced our codes at: https://github.com/IRMVLab/EfficientLO-Net.

Depth and ego-motion estimations are essential for the localization and navigation of autonomous robots and autonomous driving. Recent studies make it possible to learn the per-pixel depth and ego-motion from the unlabeled monocular video. In this paper, a novel unsupervised training framework is proposed with 3D hierarchical refinement and augmentation using explicit 3D geometry. In this framework, the depth and pose estimations are hierarchically and mutually coupled to refine the estimated pose layer by layer. The intermediate view image is proposed and synthesized by warping the pixels in an image with the estimated depth and coarse pose. Then, the residual pose transformation can be estimated from the new view image and the image of the adjacent frame to refine the coarse pose. The iterative refinement is implemented in a differentiable manner in this paper, making the whole framework optimized uniformly. Meanwhile, a new image augmentation method is proposed for the pose estimation by synthesizing a new view image, which creatively augments the pose in 3D space but gets a new augmented 2D image. The experiments on KITTI demonstrate that our depth estimation achieves state-of-the-art performance and even surpasses recent approaches that utilize other auxiliary tasks. Our visual odometry outperforms all recent unsupervised monocular learning-based methods and achieves competitive performance to the geometry-based method, ORB-SLAM2 with back-end optimization. The source codes will be released soon at:https://github.com/IRMVLab/HRANet.

3-D scene flow characterizes how the points at the current time flow to the next time in the 3-D Euclidean space, which possesses the capacity to infer autonomously the nonrigid motion of all objects in the scene. The previous methods for estimating scene flow from images have limitations, which split the holistic nature of 3-D scene flow by estimating optical flow and disparity separately. Learning 3-D scene flow from point clouds also faces the difficulties of the gap between synthesized and real data and the sparsity of LiDAR point clouds. In this article, the generated dense depth map is utilized to obtain explicit 3-D coordinates, which achieves direct learning of 3-D scene flow from 2-D images. The stability of the predicted scene flow is improved by introducing the dense nature of 2-D pixels into the 3-D space. Outliers in the generated 3-D point cloud are removed by statistical methods to weaken the impact of noisy points on the 3-D scene flow estimation task. Disparity consistency loss is proposed to achieve more effective unsupervised learning of 3-D scene flow. The proposed method of self-supervised learning of 3-D scene flow on real-world images is compared with a variety of methods for learning on the synthesized dataset and learning on LiDAR point clouds. The comparisons of multiple scene flow metrics are shown to demonstrate the effectiveness and superiority of introducing pseudo-LiDAR point cloud to scene flow estimation.

Mobile robot autonomous navigation in large-scale environments with crowded dynamic objects and static obstacles is still an essential yet challenging task. Recent works have demonstrated the potential of using deep reinforcement learning to enable autonomous navigation in crowds. However, only considering the human-robot interactions results in short-sighted and unsafe behaviors, and they typically use hand-crafted features and assume the global observation range, leading to large performance declines in large-scale crowded environments. Recent advances have shown the power of graph neural networks to learn local interactions among surrounding objects. In this paper, we consider autonomous navigation task in large-scale environments with crowded static and dynamic objects (such as humans). Particularly, local interactions among dynamic objects are learned for better-understanding their moving tendency and relational graph learning is introduced for aggregating both the object-object interactions and object-robot interactions. In addition, local observations are transformed into graphical inputs to achieve the scalability to various number of surrounding dynamic objects and various static obstacle patterns, and the globally guided reinforcement learning strategy is introduced to achieve the fixed-sized learning model even in large-scale complex environments. Simulation results validate our generalizability to various environments and advanced performance compared with existing works in large-scale crowded environments. In particular, our method with only local observations performs better than the benchmarks with global complete observability. Finally, physical robotic experiments demonstrate our effectiveness and practical applicability in real scenarios.

Multiple Object Tracking (MOT) is a significant task in autonomous driving. Nonetheless, relying on one single sensor is not robust enough, because one modality tends to fail in some challenging situations. Texture information from RGB cameras and 3D structure information from Light Detection and Ranging (LiDAR) have respective advantages under different circumstances. Therefore, feature fusion from multiple modalities contributes to the learning of discriminative features. However, it is nontrivial to achieve effective feature fusion due to the completely distinct information modality. Previous fusion methods usually fuse the top-level features after the backbones extract the features from different modalities. The feature fusion happens solely once, which limits the information interaction between different modalities. In this paper, we propose multi-scale interactive query and fusion between pixel-wise and point-wise features to obtain more discriminative features. In addition, an attention mechanism is utilized to conduct soft feature fusion between multiple pixels and points to avoid inaccurate match problems of previous single pixel-point fusion methods. We introduce PointNet++ to obtain multi-scale deep representations of point clouds and make it adaptive to our proposed interactive feature fusion between multi-scale features of images and point clouds. Through the interaction module, each modality can integrate more complementary information from the other modality. Besides, we explore the effectiveness of pre-training on each single modality and fine-tuning on the fusion-based model. Our method can achieve 90.32% MOTA and 72.44% HOTA on the KITTI benchmark and outperform other approaches without using multi-scale soft feature fusion.

In this paper, we propose a novel network, Fusion-Net, which can estimate the extrinsic calibration matrix between LiDAR and a monocular RGB camera with high accuracy and robustness. FusionNet is a coarse-to-fine method, providing an online and end-to-end solution that can automatically detect and correct the decalibration without any specially designed targets or environments. First, the network applies deep-learning-based technologies to extract the features of LiDAR point clouds and RGB images. Then a novel method is adopted to fuse the features got from different sensors by projecting LiDAR features onto RGB feature maps, searching for the RGB features with the projected points as centers and concatenating the extracted RGB features with LiDAR features. To increase the accuracy, we apply a coarse-to-fine method in the network, by transforming LiDAR points and estimating the extrinsic calibration matrices from the coarse scale to the fine scale. The network is trained on random artificial decalibration matrices. Compared to existing approaches, our method doesn't need to train additional iterative networks, but it can also adapt to different ranges of decalibration.

The semantic segmentation of point clouds is an important part of the environment perception for robots. However, it is difficult to directly adopt the traditional 3-D convolution kernel to extract features from raw 3-D point clouds because of the unstructured property of point clouds. In this article, a spherical interpolated convolution operator is proposed to replace the traditional grid-shaped 3-D convolution operator. In addition, this article analyzes the defect of point cloud interpolation methods based on the distance as the interpolation weight and proposes the self-learned distance-feature density by combining the distance and the feature correlation. The proposed method makes the feature extraction of the spherical interpolated convolution network more rational and effective. The effectiveness of the proposed network is demonstrated on the 3-D semantic segmentation task of point clouds. Experiments show that the proposed method achieves good performance on the ScanNet dataset and Paris-Lille-3D dataset. The comparison experiments with the traditional grid-shaped 3-D convolution operator demonstrated that the newly proposed feature extraction operator improves the accuracy of the network and reduces the parameters of the network. The source codes will be released on https://github.com/IRMVLab/SIConv.

Optical flow estimation is a fundamental problem of computer vision and has many applications in the fields of robot learning and autonomous driving. This paper reveals novel geometric principles of optical flow based on the insight and detailed definition of non-occlusion. Then, two novel loss functions are proposed for the unsupervised learning of optical flow based on the geometric principles of non-occlusion. Specifically, after the occlusion part of the images are masked, the flowing process of pixel coordinate points is carefully considered and geometric constraints are conducted based on the geometric principles of optical flow. First, the optical flow vectors of neighboring pixel coordinate points in the first frame will not intersect during the pixel displacement to the second frame. Secondly, when the cluster containing adjacent four pixel coordinate points in the first frame moves to the second frame, no other pixel coordinate points will flow into the quadrilateral formed by them. According to the two geometrical constraints, the optical flow non-intersection loss and the optical flow non-blocking loss in the non-occlusion regions are proposed. Two loss functions punish the irregular and inexact optical flows in the non-occlusion regions. The experiments on datasets demonstrated that the proposed unsupervised losses of optical flow based on the geometric principles in non-occlusion regions make the estimated optical flow more refined in detail, and improve the performance of unsupervised learning of optical flow. In addition, the experiments training on synthetic data and evaluating on real data show that the generalization ability of optical flow network is improved by our proposed unsupervised approach. Our codes are open sourced at:https://github.com/IRMVLab/NccFlow.

In autonomous driving, monocular sequences contain lots of information. Monocular depth estimation, camera ego-motion estimation and optical flow estimation in consecutive frames are high-profile concerns recently. By analyzing tasks above, pixels in the middle frame are modeled into three parts: the rigid region, the non-rigid region, and the occluded region. In joint unsupervised training of depth and pose, we can segment the occluded region explicitly. The occlusion information is used in unsupervised learning of depth, pose and optical flow, as the image reconstructed by depth-pose and optical flow will be invalid in occluded regions. A less-than-mean mask is designed to further exclude the mismatched pixels interfered with by motion or illumination change in the training of depth and pose networks. This method is also used to exclude some trivial mismatched pixels in the training of the optical flow network. Maximum normalization is proposed for depth smoothness term to restrain depth degradation in textureless regions. In the occluded region, as depth and camera motion can provide more reliable motion estimation, they can be used to instruct unsupervised learning of optical flow. Our experiments in KITTI dataset demonstrate that the model based on three regions, full and explicit segmentation of the occlusion region, the rigid region, and the non-rigid region with corresponding unsupervised losses can improve performance on three tasks significantly. The source code is available at:https://github.com/guangmingw/DOPlearning.

A novel 3D point cloud learning model for deep LiDAR odometry, named PWCLO-Net, using hierarchical embedding mask optimization is proposed in this paper. In this model, the Pyramid, Warping, and Cost volume (PWC) structure for the LiDAR odometry task is built to refine the estimated pose in a coarse-to-fine approach hierarchically. An attentive cost volume is built to associate two point clouds and obtain embedding motion patterns. Then, a novel trainable embedding mask is proposed to weigh the local motion patterns of all points to regress the overall pose and filter outlier points. The estimated current pose is used to warp the first point cloud to bridge the distance to the second point cloud, and then the cost volume of the residual motion is built. At the same time, the embedding mask is optimized hierarchically from coarse to fine to obtain more accurate filtering information for pose refinement. The trainable pose warp-refinement process is iteratively used to make the pose estimation more robust for outliers. The superior performance and effectiveness of our LiDAR odometry model are demonstrated on KITTI odometry dataset. Our method outperforms all recent learning-based methods and outperforms the geometry-based approach, LOAM with mapping optimization, on most sequences of KITTI odometry dataset. Our source codes will be released on https://github.com/IRMVLab/PWCLONet.

Scene flow represents the motion of points in the 3D space, which is the counterpart of the optical flow that represents the motion of pixels in the 2D image. However, it is difficult to obtain the ground truth of scene flow in the real scenes, and recent studies are based on synthetic data for training. Therefore, how to train a scene flow network with unsupervised methods based on real-world data shows crucial significance. A novel unsupervised learning method for scene flow is proposed in this paper, which utilizes the images of two consecutive frames taken by monocular camera without the ground truth of scene flow for training. Our method realizes the goal that training scene flow network with real-world data, which bridges the gap between training data and test data and broadens the scope of available data for training. Unsupervised learning of scene flow in this paper mainly consists of two parts: (i) depth estimation and camera pose estimation, and (ii) scene flow estimation based on four different loss functions. Depth estimation and camera pose estimation obtain the depth maps and camera pose between two consecutive frames, which provide further information for the next scene flow estimation. After that, we used depth consistency loss, dynamic-static consistency loss, Chamfer loss, and Laplacian regularization loss to carry out unsupervised training of the scene flow network. To our knowledge, this is the first paper that realizes the unsupervised learning of 3D scene flow from monocular camera. The experiment results on KITTI show that our method for unsupervised learning of scene flow meets great performance compared to traditional methods Iterative Closest Point (ICP) and Fast Global Registration (FGR). The source code is available at: https://github.com/IRMVLab/3DUnMonoFlow.

Scene flow represents the 3D motion of every point in the dynamic environments. Like the optical flow that represents the motion of pixels in 2D images, 3D motion representation of scene flow benefits many applications, such as autonomous driving and service robot. This paper studies the problem of scene flow estimation from two consecutive 3D point clouds. In this paper, a novel hierarchical neural network with double attention is proposed for learning the correlation of point features in adjacent frames and refining scene flow from coarse to fine layer by layer. The proposed network has a new more-for-less hierarchical architecture. The more-for-less means that the number of input points is greater than the number of output points for scene flow estimation, which brings more input information and balances the precision and resource consumption. In this hierarchical architecture, scene flow of different levels is generated and supervised respectively. A novel attentive embedding module is introduced to aggregate the features of adjacent points using a double attention method in a patch-to-patch manner. The proper layers for flow embedding and flow supervision are carefully considered in our network designment. Experiments show that the proposed network outperforms the state-of-the-art performance of 3D scene flow estimation on the FlyingThings3D and KITTI Scene Flow 2015 datasets. We also apply the proposed network to the realistic LiDAR odometry task, which is a key problem in autonomous driving. The experiment results demonstrate that our proposed network can outperform the ICP-based method and shows good practical application ability. The source codes will be released on https://github.com/IRMVLab/HALFlow.

A new unsupervised learning method of depth and ego-motion using multiple masks from monocular video is proposed in this paper. The depth estimation network and the ego-motion estimation network are trained according to the constraints of depth and ego-motion without truth values. The main contribution of our method is to carefully consider the occlusion of the pixels generated when the adjacent frames are projected to each other, and the blank problem generated in the projection target imaging plane. Two fine masks are designed to solve most of the image pixel mismatch caused by the movement of the camera. In addition, some relatively rare circumstances are considered, and repeated masking is proposed. To some extent, the method is to use a geometric relationship to filter the mismatched pixels for training, making unsupervised learning more efficient and accurate. The experiments on KITTI dataset show our method achieves good performance in terms of depth and ego-motion. The generalization capability of our method is demonstrated by training on the low-quality uncalibrated bike video dataset and evaluating on KITTI dataset, and the results are still good.