Depth completion is a popular research direction in the field of depth estimation. The fusion of color and depth features is the critical challenge in this task, mainly due to the asymmetry between the rich scene details in color images and the sparse pixels in depth maps. To tackle this issue, we design an efficient Gated Cross-Attention Network that propagates confidence via a gating mechanism, simultaneously extracting and refining key information in both color and depth branches to achieve local spatial feature fusion. Additionally, we incorporate a Transformer-based attention network in low-dimensional space to effectively fuse global features and increase the network’s receptive field. At the same time, we use the Ray Tune mechanism with the AsyncHyperBandScheduler and the HyperOptSearch algorithm to automatically search for the optimal number of module iterations, which also allows us to achieve performance comparable to state-of-the-art methods. We conduct experiments on both indoor and outdoor scene datasets. Our fast network ranked first among real-time methods (below 30ms and 100ms), and our accurate network ranked first among all methods on the KITTI official website at the time of submission.

Diffusion models optimized via variational inference (VI) have emerged as a promising tool for generating samples from unnormalized target densities. These models create samples by simulating a stochastic differential equation, starting from a simple, tractable prior, typically a Gaussian distribution. However, when the support of this prior differs greatly from that of the target distribution, diffusion models often struggle to explore effectively or suffer from large discretization errors. Moreover, learning the prior distribution can lead to mode-collapse, exacerbated by the mode-seeking nature of reverse Kullback-Leibler divergence commonly used in VI. To address these challenges, we propose end-to-end learnable Gaussian mixture priors (GMPs). GMPs offer improved control over exploration, adaptability to target support, and increased expressiveness to counteract mode collapse. We further leverage the structure of mixture models by proposing a strategy to iteratively refine the model through the addition of mixture components during training. Our experimental results demonstrate significant performance improvements across a diverse range of real-world and synthetic benchmark problems when using GMPs without requiring additional target evaluations.

Robotic manipulation systems benefit from complementary sensing modalities, where each provides unique environmental information. Point clouds capture detailed geometric structure, while RGB images provide rich semantic context. Current point cloud methods struggle to capture fine-grained detail, especially for complex tasks, which RGB methods lack geometric awareness, which hinders their precision and generalization. We introduce PointMapPolicy, a novel approach that conditions diffusion policies on structured grids of points without downsampling. The resulting data type makes it easier to extract shape and spatial relationships from observations, and can be transformed between reference frames. Yet due to their structure in a regular grid, we enable the use of established computer vision techniques directly to 3D data. Using xLSTM as a backbone, our model efficiently fuses the point maps with RGB data for enhanced multi-modal perception. Through extensive experiments on the RoboCasa and CALVIN benchmarks and real robot evaluations, we demonstrate that our method achieves state-of-the-art performance across diverse manipulation tasks. The overview and demos are available on our project page: https://point-map.github.io/Point-Map/

We present the B-spline Encoded Action Sequence Tokenizer (BEAST), a novel action tokenizer that encodes action sequences into compact discrete or continuous tokens using B-splines. In contrast to existing action tokenizers based on vector quantization or byte pair encoding, BEAST requires no separate tokenizer training and consistently produces tokens of uniform length, enabling fast action sequence generation via parallel decoding. Leveraging our B-spline formulation, BEAST inherently ensures generating smooth trajectories without discontinuities between adjacent segments. We extensively evaluate BEAST by integrating it with three distinct model architectures: a Variational Autoencoder (VAE) with continuous tokens, a decoder-only Transformer with discrete tokens, and Florence-2, a pretrained Vision-Language Model with an encoder-decoder architecture, demonstrating BEAST's compatibility and scalability with large pretrained models. We evaluate BEAST across three established benchmarks consisting of 166 simulated tasks and on three distinct robot settings with a total of 8 real-world tasks. Experimental results demonstrate that BEAST (i) significantly reduces both training and inference computational costs, and (ii) consistently generates smooth, high-frequency control signals suitable for continuous control tasks while (iii) reliably achieves competitive task success rates compared to state-of-the-art methods.

Future versatile robots need the ability to learn new tasks and behaviors from demonstrations. Recent advances in virtual and augmented reality position these technologies as great candidates for the efficient and intuitive collection of large sets of demonstrations. While there are different possible approaches to control a virtual robot there has not yet been an evaluation of these control interfaces in regards to their efficiency and intuitiveness. These characteristics become particularly important when working with non-expert users and complex manipulation tasks. To this end, this work investigates five different interfaces to control a virtual robot in a comprehensive user study across various virtualized tasks in an AR setting. These interfaces include Hand Tracking, Virtual Kinesthetic Teaching, Gamepad and Motion Controller. Additionally, this work introduces Kinesthetic Teaching as a novel interface to control virtual robots in AR settings, where the virtual robot mimics the movement of a real robot manipulated by the user. This study reveals valuable insights into their usability and effectiveness. It shows that the proposed Kinesthetic Teaching interface significantly outperforms other interfaces in both objective and subjective metrics based on success rate, task completeness, and completion time and User Experience Questionnaires (UEQ+).

Imitation learning with human data has demonstrated remarkable success in teaching robots in a wide range of skills. However, the inherent diversity in human behavior leads to the emergence of multi-modal data distributions, thereby presenting a formidable challenge for existing imitation learning algorithms. Quantifying a model's capacity to capture and replicate this diversity effectively is still an open problem. In this work, we introduce simulation benchmark environments and the corresponding *Datasets with Diverse human Demonstrations for Imitation Learning (D3IL)*, designed explicitly to evaluate a model's ability to learn multi-modal behavior. Our environments are designed to involve multiple sub-tasks that need to be solved, consider manipulation of multiple objects which increases the diversity of the behavior and can only be solved by policies that rely on closed loop sensory feedback. Other available datasets are missing at least one of these challenging properties. To address the challenge of diversity quantification, we introduce tractable metrics that provide valuable insights into a model's ability to acquire and reproduce diverse behaviors. These metrics offer a practical means to assess the robustness and versatility of imitation learning algorithms. Furthermore, we conduct a thorough evaluation of state-of-the-art methods on the proposed task suite. This evaluation serves as a benchmark for assessing their capability to learn diverse behaviors. Our findings shed light on the effectiveness of these methods in tackling the intricate problem of capturing and generalizing multi-modal human behaviors, offering a valuable reference for the design of future imitation learning algorithms.

Monte Carlo methods, Variational Inference, and their combinations play a pivotal role in sampling from intractable probability distributions. However, current studies lack a unified evaluation framework, relying on disparate performance measures and limited method comparisons across diverse tasks, complicating the assessment of progress and hindering the decision-making of practitioners. In response to these challenges, our work introduces a benchmark that evaluates sampling methods using a standardized task suite and a broad range of performance criteria. Moreover, we study existing metrics for quantifying mode collapse and introduce novel metrics for this purpose. Our findings provide insights into strengths and weaknesses of existing sampling methods, serving as a valuable reference for future developments.

This work introduces Variational Diffusion Distillation (VDD), a novel method that distills denoising diffusion policies into Mixtures of Experts (MoE) through variational inference. Diffusion Models are the current state-of-the-art in generative modeling due to their exceptional ability to accurately learn and represent complex, multi-modal distributions. This ability allows Diffusion Models to replicate the inherent diversity in human behavior, making them the preferred models in behavior learning such as Learning from Human Demonstrations (LfD). However, diffusion models come with some drawbacks, including the intractability of likelihoods and long inference times due to their iterative sampling process. The inference times, in particular, pose a significant challenge to real-time applications such as robot control. In contrast, MoEs effectively address the aforementioned issues while retaining the ability to represent complex distributions but are notoriously difficult to train. VDD is the first method that distills pre-trained diffusion models into MoE models, and hence, combines the expressiveness of Diffusion Models with the benefits of Mixture Models. Specifically, VDD leverages a decompositional upper bound of the variational objective that allows the training of each expert separately, resulting in a robust optimization scheme for MoEs. VDD demonstrates across nine complex behavior learning tasks, that it is able to: i) accurately distill complex distributions learned by the diffusion model, ii) outperform existing state-of-the-art distillation methods, and iii) surpass conventional methods for training MoE. The code and videos are available at https://intuitive-robots.github.io/vdd-website.

Imitation learning uses data for training policies to solve complex tasks. However, when the training data is collected from human demonstrators, it often leads to multimodal distributions because of the variability in human actions. Most imitation learning methods rely on a maximum likelihood (ML) objective to learn a parameterized policy, but this can result in suboptimal or unsafe behavior due to the mode-averaging property of the ML objective. In this work, we propose Information Maximizing Curriculum, a curriculum-based approach that assigns a weight to each data point and encourages the model to specialize in the data it can represent, effectively mitigating the mode-averaging problem by allowing the model to ignore data from modes it cannot represent. To cover all modes and thus, enable versatile behavior, we extend our approach to a mixture of experts (MoE) policy, where each mixture component selects its own subset of the training data for learning. A novel, maximum entropy-based objective is proposed to achieve full coverage of the dataset, thereby enabling the policy to encompass all modes within the data distribution. We demonstrate the effectiveness of our approach on complex simulated control tasks using versatile human demonstrations, achieving superior performance compared to state-of-the-art methods.

Accurately predicting the possible behaviors of traffic participants is an essential capability for autonomous vehicles. Since autonomous vehicles need to navigate in dynamically changing environments, they are expected to make accurate predictions regardless of where they are and what driving circumstances they encountered. Therefore, generalization capability to unseen domains is crucial for prediction models when autonomous vehicles are deployed in the real world. In this paper, we aim to address the domain generalization problem for vehicle intention prediction tasks and a causal-based time series domain generalization (CTSDG) model is proposed. We construct a structural causal model for vehicle intention prediction tasks to learn an invariant representation of input driving data for domain generalization. We further integrate a recurrent latent variable model into our structural causal model to better capture temporal latent dependencies from time-series input data. The effectiveness of our approach is evaluated via real-world driving data. We demonstrate that our proposed method has consistent improvement on prediction accuracy compared to other state-of-the-art domain generalization and behavior prediction methods.

Stereo matching has attracted much attention in recent years. Traditional methods can quickly generate a disparity result, but the accuracy is low. On the contrary, methods based on neural networks can achieve a high accuracy level, but they are difficult to reach the real-time level. Therefore, this paper presents MCDRNet, which combines traditional methods with neural networks to achieve real-time and accurate stereo matching results. Concretely, our network first generates a rough disparity map based on the traditional ADCensus algorithm. Then we design a novel Multi-Scale Cascade Network to refine the disparity map from coarse to fine. We evaluate our best-trained model on the KITTI official website. The results show that our network is much faster than most current top-performing methods(31×than CSPN, 56×than GANet, etc.). Meanwhile, it is more accurate than traditional stereo methods(SGM, SPS-St) and other fast 2D convolution networks(Fast DS-CS, DispNetC, etc.), demonstrating the rationalities and feasibilities of our method.

Stereo matching is essential for robot navigation. However, the accuracy of current widely used traditional methods is low, while methods based on CNN need expensive computational cost and running time. This is because different cost volumes play a crucial role in balancing speed and accuracy. Thus we propose MSCVNet, which combines traditional methods and neural networks to improve the quality of cost volume. Concretely, our network first generates multiple 3D cost volumes with different resolutions and then uses 2D convolutions to construct a novel cascade hourglass network for cost aggregation. Meanwhile, we design an algorithm to distinguish and calculate the loss for discontinuous areas of the disparity result. According to the KITTI official website, our network is much faster than most top-performing methods (24than CSPN, 44than GANet, etc.). Meanwhile, compared to traditional methods (SPS-St, SGM) and other real-time stereo matching networks (Fast DS-CS, DispNetC, and RTSNet, etc.), our network achieves a big improvement in accuracy, demonstrating the feasibility and capability of the proposed method.

Liver cancer is one of the cancers with the highest mortality. In order to help doctors diagnose and treat liver lesion, an automatic liver segmentation model is urgently needed due to manually segmentation is time-consuming and error-prone. In this paper, we propose a nested attention-aware segmentation network, named Attention UNet++. Our proposed method has a deep supervised encoder-decoder architecture and a redesigned dense skip connection. Attention UNet++ introduces attention mechanism between nested convolutional blocks so that the features extracted at different levels can be merged with a task-related selection. Besides, due to the introduction of deep supervision, the prediction speed of the pruned network is accelerated at the cost of modest performance degradation. We evaluated proposed model on MICCAI 2017 Liver Tumor Segmentation (LiTS) Challenge Dataset. Attention UNet++ achieved very competitive performance for liver segmentation.

Object detection has long been an important issue in the discipline of scene understanding. Existing researches mainly focus on the object itself, ignoring its surrounding environment. In fact, the surrounding environment provides abundant information to help detectors classify and locate objects. This paper proposes CRPDet, viz. CenterRepp Detector, a framework for object detection. The main function of CRPDet is accomplished by the CenterRepp module, which takes into account the surrounding environment by predicting the distribution of the central representative points. CenterRepp converts labeled object frames into the mean and standard variance of the sampling points' distribution. This helps increase the receptive field of objects, breaking the limitation of object frames. CenterRepp defines a position-fixed center point with significant weights, avoiding to sample all points in the surroundings. Experiments on the COCO test-dev detection benchmark demonstrates that our proposed CRPDet has comparable performance with state-of-the-art detectors, achieving 39.4 mAP with 51 FPS tested under single size input.

In this paper, we propose a modified CenterNet to complete the defect detection of Sanitary Ceramics. Generally, visual quality inspection is rather important during the productive process of Sanitary Ceramics and it is nearly impossible to inspect the massive images by hand. Consequently, it is necessary to devise an accurate and real-time system to process the data. However, due to the varied shapes and backgrounds of ceramics, conventional computer vision methods are usually not robust to all those variables. Detectors based on Deep Learning start to be adopted in recent years, but most algorithms require some carefully devised anchor boxes and post-processing methods, which also bring more computational costs. Here we decide to take advantage of the anchor-free model, CenterNet. We change the main structure to fit our own data and introduce an extra branch with shallow layers to strengthen the feature representation. The results have shown the great power of this model. Without even any post-processing methods, our model achieves a result of 96.16 AP on the established dataset.