Asako Kanezaki

Kei Ota, Devesh K. Jha, Krishna Murthy Jatavallabhula et al. · mit

Precise perception of contact interactions is essential for fine-grained manipulation skills for robots. In this paper, we present the design of feedback skills for robots that must learn to stack complex-shaped objects on top of each other (see Fig.1). To design such a system, a robot should be able to reason about the stability of placement from very gentle contact interactions. Our results demonstrate that it is possible to infer the stability of object placement based on tactile readings during contact formation between the object and its environment. In particular, we estimate the contact patch between a grasped object and its environment using force and tactile observations to estimate the stability of the object during a contact formation. The contact patch could be used to estimate the stability of the object upon release of the grasp. The proposed method is demonstrated in various pairs of objects that are used in a very popular board game.

Hao Zheng, Runqi Wang, Jianzhuang Liu et al.

The conventional few-shot classification aims at learning a model on a large labeled base dataset and rapidly adapting to a target dataset that is from the same distribution as the base dataset. However, in practice, the base and the target datasets of few-shot classification are usually from different domains, which is the problem of cross-domain few-shot classification. We tackle this problem by making a small proportion of unlabeled images in the target domain accessible in the training stage. In this setup, even though the base data are sufficient and labeled, the large domain shift still makes transferring the knowledge from the base dataset difficult. We meticulously design a cross-level knowledge distillation method, which can strengthen the ability of the model to extract more discriminative features in the target dataset by guiding the network's shallow layers to learn higher-level information. Furthermore, in order to alleviate the overfitting in the evaluation stage, we propose a feature denoising operation which can reduce the feature redundancy and mitigate overfitting. Our approach can surpass the previous state-of-the-art method, Dynamic-Distillation, by 5.44% on 1-shot and 1.37% on 5-shot classification tasks on average in the BSCD-FSL benchmark. The implementation code will be available at https://github.com/jarucezh/cldfd.

Kei Ota, Hsiao-Yu Tung, Kevin A. Smith et al.

The world is filled with articulated objects that are difficult to determine how to use from vision alone, e.g., a door might open inwards or outwards. Humans handle these objects with strategic trial-and-error: first pushing a door then pulling if that doesn't work. We enable these capabilities in autonomous agents by proposing "Hypothesize, Simulate, Act, Update, and Repeat" (H-SAUR), a probabilistic generative framework that simultaneously generates a distribution of hypotheses about how objects articulate given input observations, captures certainty over hypotheses over time, and infer plausible actions for exploration and goal-conditioned manipulation. We compare our model with existing work in manipulating objects after a handful of exploration actions, on the PartNet-Mobility dataset. We further propose a novel PuzzleBoxes benchmark that contains locked boxes that require multiple steps to solve. We show that the proposed model significantly outperforms the current state-of-the-art articulated object manipulation framework, despite using zero training data. We further improve the test-time efficiency of H-SAUR by integrating a learned prior from learning-based vision models.

Rui Fukushima, Kei Ota, Asako Kanezaki et al.

This paper presents a reinforcement learning method for object goal navigation (ObjNav) where an agent navigates in 3D indoor environments to reach a target object based on long-term observations of objects and scenes. To this end, we propose Object Memory Transformer (OMT) that consists of two key ideas: 1) Object-Scene Memory (OSM) that enables to store long-term scenes and object semantics, and 2) Transformer that attends to salient objects in the sequence of previously observed scenes and objects stored in OSM. This mechanism allows the agent to efficiently navigate in the indoor environment without prior knowledge about the environments, such as topological maps or 3D meshes. To the best of our knowledge, this is the first work that uses a long-term memory of object semantics in a goal-oriented navigation task. Experimental results conducted on the AI2-THOR dataset show that OMT outperforms previous approaches in navigating in unknown environments. In particular, we show that utilizing the long-term object semantics information improves the efficiency of navigation.

Haru Kondoh, Asako Kanezaki

Over the past few years, there has been a great deal of research on navigation tasks in indoor environments using deep reinforcement learning agents. Most of these tasks use only visual information in the form of first-person images to navigate to a single goal. More recently, tasks that simultaneously use visual and auditory information to navigate to the sound source and even navigation tasks with multiple goals instead of one have been proposed. However, there has been no proposal for a generalized navigation task combining these two types of tasks and using both visual and auditory information in a situation where multiple sound sources are goals. In this paper, we propose a new framework for this generalized task: multi-goal audio-visual navigation. We first define the task in detail, and then we investigate the difficulty of the multi-goal audio-visual navigation task relative to the current navigation tasks by conducting experiments in various situations. The research shows that multi-goal audio-visual navigation has the difficulty of the implicit need to separate the sources of sound. Next, to mitigate the difficulties in this new task, we propose a method named sound direction map (SDM), which dynamically localizes multiple sound sources in a learning-based manner while making use of past memories. Experimental results show that the use of SDM significantly improves the performance of multiple baseline methods, regardless of the number of goals.

Yuchen Che, Ryo Furukawa, Asako Kanezaki

Category-level articulated object pose estimation focuses on the pose estimation of unknown articulated objects within known categories. Despite its significance, this task remains challenging due to the varying shapes and poses of objects, expensive dataset annotation costs, and complex real-world environments. In this paper, we propose a novel self-supervised approach that leverages a single-frame point cloud to solve this task. Our model consistently generates reconstruction with a canonical pose and joint state for the entire input object, and it estimates object-level poses that reduce overall pose variance and part-level poses that align each part of the input with its corresponding part of the reconstruction. Experimental results demonstrate that our approach significantly outperforms previous self-supervised methods and is comparable to the state-of-the-art supervised methods. To assess the performance of our model in real-world scenarios, we also introduce a new real-world articulated object benchmark dataset.

Nanami Kotani, Asako Kanezaki

One of the intuitive instruction methods in robot navigation is a pointing gesture. In this study, we propose a method using an omnidirectional camera to eliminate the user/object position constraint and the left/right constraint of the pointing arm. Although the accuracy of skeleton and object detection is low due to the high distortion of equirectangular images, the proposed method enables highly accurate estimation by repeatedly extracting regions of interest from the equirectangular image and projecting them onto perspective images. Furthermore, we found that training the likelihood of the target object in machine learning further improves the estimation accuracy.

Ayaka Nanri, Klara Reichard, Mert Kiray et al.

Recent advances in 3D datasets and multimodal models have greatly improved natural language 3D scene understanding. However, most 3D referring segmentation methods do not explicitly represent the observer viewpoint, making spatial relations such as "left," "right," "front," and "behind" ambiguous and difficult to evaluate. We introduce a viewpoint-aware 3D referring segmentation dataset containing 220k benchmark samples, and scalable to tens of millions of viewpoint-conditioned samples through dense viewpoint sampling. In this dataset, target objects can only be identified through observer-centric spatial relations, making viewpoint-conditioned grounding necessary. We construct the benchmark by leveraging camera poses to automatically annotate observer-centric relations (left/right, front/behind) together with viewpoint-independent relations (above/under). Using this benchmark, we evaluate several existing 3D large multimodal models in a zero-shot setting and find that current models struggle with viewpoint-dependent spatial instructions. We further study how explicit viewpoint information can be incorporated into 3D large multimodal models. We introduce a viewpoint representation that encodes camera poses and conditions the model on the observation viewpoint, improving segmentation accuracy on viewpoint-dependent relations and increasing mIoU from 0.30 to 0.47 compared to a model without viewpoint conditioning. The dataset, code, and trained models will be made publicly available upon acceptance.

Asako Kanezaki, Yasuyuki Matsushita, Yoshifumi Nishida

We propose a Convolutional Neural Network (CNN)-based model "RotationNet," which takes multi-view images of an object as input and jointly estimates its pose and object category. Unlike previous approaches that use known viewpoint labels for training, our method treats the viewpoint labels as latent variables, which are learned in an unsupervised manner during the training using an unaligned object dataset. RotationNet is designed to use only a partial set of multi-view images for inference, and this property makes it useful in practical scenarios where only partial views are available. Moreover, our pose alignment strategy enables one to obtain view-specific feature representations shared across classes, which is important to maintain high accuracy in both object categorization and pose estimation. Effectiveness of RotationNet is demonstrated by its superior performance to the state-of-the-art methods of 3D object classification on 10- and 40-class ModelNet datasets. We also show that RotationNet, even trained without known poses, achieves the state-of-the-art performance on an object pose estimation dataset. The code is available on https://github.com/kanezaki/rotationnet

Leyuan Sun, Asako Kanezaki, Guillaume Caron et al.

Object-goal navigation is a crucial engineering task for the community of embodied navigation; it involves navigating to an instance of a specified object category within unseen environments. Although extensive investigations have been conducted on both end-to-end and modular-based, data-driven approaches, fully enabling an agent to comprehend the environment through perceptual knowledge and perform object-goal navigation as efficiently as humans remains a significant challenge. Recently, large language models have shown potential in this task, thanks to their powerful capabilities for knowledge extraction and integration. In this study, we propose a data-driven, modular-based approach, trained on a dataset that incorporates common-sense knowledge of object-to-room relationships extracted from a large language model. We utilize the multi-channel Swin-Unet architecture to conduct multi-task learning incorporating with multimodal inputs. The results in the Habitat simulator demonstrate that our framework outperforms the baseline by an average of 10.6% in the efficiency metric, Success weighted by Path Length (SPL). The real-world demonstration shows that the proposed approach can efficiently conduct this task by traversing several rooms. For more details and real-world demonstrations, please check our project webpage (https://sunleyuan.github.io/ObjectNav).

Haru Kondoh, Asako Kanezaki

The field of multimodal robot navigation in indoor environments has garnered significant attention in recent years. However, as tasks and methods become more advanced, the action decision systems tend to become more complex and operate as black-boxes. For a reliable system, the ability to explain or describe its decisions is crucial; however, there tends to be a trade-off in that explainable systems can not outperform non-explainable systems in terms of performance. In this paper, we propose incorporating the task of describing actions in language into the reinforcement learning of navigation as an auxiliary task. Existing studies have found it difficult to incorporate describing actions into reinforcement learning due to the absence of ground-truth data. We address this issue by leveraging knowledge distillation from pre-trained description generation models, such as vision-language models. We comprehensively evaluate our approach across various navigation tasks, demonstrating that it can describe actions while attaining high navigation performance. Furthermore, it achieves state-of-the-art performance in the particularly challenging multimodal navigation task of semantic audio-visual navigation.

Jiawei Jiang, Kei Ota, Devesh K. Jha et al.

Incorporating additional sensory modalities such as tactile and audio into foundational robotic models poses significant challenges due to the curse of dimensionality. This work addresses this issue through modality selection. We propose a cross-modality attention (CMA) mechanism to identify and selectively utilize the modalities that are most informative for action generation at each timestep. Furthermore, we extend the application of CMA to segment primitive skills from expert demonstrations and leverage this segmentation to train a hierarchical policy capable of solving long-horizon, contact-rich manipulation tasks.

Kuanting Wu, Kei Ota, Asako Kanezaki

Video Diffusion Models (VDMs) can generate high-quality videos, but often struggle with producing temporally coherent motion. Optical flow supervision is a promising approach to address this, with prior works commonly employing warping-based strategies that avoid explicit flow matching. In this work, we explore an alternative formulation, FlowLoss, which directly compares flow fields extracted from generated and ground-truth videos. To account for the unreliability of flow estimation under high-noise conditions in diffusion, we propose a noise-aware weighting scheme that modulates the flow loss across denoising steps. Experiments on robotic video datasets suggest that FlowLoss improves motion stability and accelerates convergence in early training stages. Our findings offer practical insights for incorporating motion-based supervision into noise-conditioned generative models.

Masaru Yajima, Kei Ota, Asako Kanezaki et al.

Achieving zero-shot peg insertion, where inserting an arbitrary peg into an unseen hole without task-specific training, remains a fundamental challenge in robotics. This task demands a highly generalizable perception system capable of detecting potential holes, selecting the correct mating hole from multiple candidates, estimating its precise pose, and executing insertion despite uncertainties. While learning-based methods have been applied to peg insertion, they often fail to generalize beyond the specific peg-hole pairs encountered during training. Recent advancements in Vision-Language Models (VLMs) offer a promising alternative, leveraging large-scale datasets to enable robust generalization across diverse tasks. Inspired by their success, we introduce a novel zero-shot peg insertion framework that utilizes a VLM to identify mating holes and estimate their poses without prior knowledge of their geometry. Extensive experiments demonstrate that our method achieves 90.2% accuracy, significantly outperforming baselines in identifying the correct mating hole across a wide range of previously unseen peg-hole pairs, including 3D-printed objects, toy puzzles, and industrial connectors. Furthermore, we validate the effectiveness of our approach in a real-world connector insertion task on a backpanel of a PC, where our system successfully detects holes, identifies the correct mating hole, estimates its pose, and completes the insertion with a success rate of 88.3%. These results highlight the potential of VLM-driven zero-shot reasoning for enabling robust and generalizable robotic assembly.

Koji Takeda, Kanji Tanaka, Yoshimasa Nakamura et al.

In everyday indoor navigation, robots often needto detect non-distinctive small-change objects (e.g., stationery,lost items, and junk, etc.) to maintain domain knowledge. Thisis most relevant to ground-view change detection (GVCD), a recently emerging research area in the field of computer vision.However, these existing techniques rely on high-quality class-specific object priors to regularize a change detector modelthat cannot be applied to semantically nondistinctive smallobjects. To address ill-posedness, in this study, we explorethe concept of degree-of-ill-posedness (DoI) from the newperspective of GVCD, aiming to improve both passive and activevision. This novel DoI problem is highly domain-dependent,and manually collecting fine-grained annotated training datais expensive. To regularize this problem, we apply the conceptof self-supervised learning to achieve efficient DoI estimationscheme and investigate its generalization to diverse datasets.Specifically, we tackle the challenging issue of obtaining self-supervision cues for semantically non-distinctive unseen smallobjects and show that novel "oversegmentation cues" from openvocabulary semantic segmentation can be effectively exploited.When applied to diverse real datasets, the proposed DoI modelcan boost state-of-the-art change detection models, and it showsstable and consistent improvements when evaluated on real-world datasets.

Keisuke Okumura, Ryo Yonetani, Mai Nishimura et al.

Multi-agent path planning (MAPP) in continuous spaces is a challenging problem with significant practical importance. One promising approach is to first construct graphs approximating the spaces, called roadmaps, and then apply multi-agent pathfinding (MAPF) algorithms to derive a set of conflict-free paths. While conventional studies have utilized roadmap construction methods developed for single-agent planning, it remains largely unexplored how we can construct roadmaps that work effectively for multiple agents. To this end, we propose a novel concept of roadmaps called cooperative timed roadmaps (CTRMs). CTRMs enable each agent to focus on its important locations around potential solution paths in a way that considers the behavior of other agents to avoid inter-agent collisions (i.e., "cooperative"), while being augmented in the time direction to make it easy to derive a "timed" solution path. To construct CTRMs, we developed a machine-learning approach that learns a generative model from a collection of relevant problem instances and plausible solutions and then uses the learned model to sample the vertices of CTRMs for new, previously unseen problem instances. Our empirical evaluation revealed that the use of CTRMs significantly reduced the planning effort with acceptable overheads while maintaining a success rate and solution quality comparable to conventional roadmap construction approaches.

Hana Hoshino, Kei Ota, Asako Kanezaki et al.

Inverse Reinforcement Learning (IRL) is attractive in scenarios where reward engineering can be tedious. However, prior IRL algorithms use on-policy transitions, which require intensive sampling from the current policy for stable and optimal performance. This limits IRL applications in the real world, where environment interactions can become highly expensive. To tackle this problem, we present Off-Policy Inverse Reinforcement Learning (OPIRL), which (1) adopts off-policy data distribution instead of on-policy and enables significant reduction of the number of interactions with the environment, (2) learns a stationary reward function that is transferable with high generalization capabilities on changing dynamics, and (3) leverages mode-covering behavior for faster convergence. We demonstrate that our method is considerably more sample efficient and generalizes to novel environments through the experiments. Our method achieves better or comparable results on policy performance baselines with significantly fewer interactions. Furthermore, we empirically show that the recovered reward function generalizes to different tasks where prior arts are prone to fail.

Kei Ota, Devesh K. Jha, Asako Kanezaki

The success of deep learning in the computer vision and natural language processing communities can be attributed to training of very deep neural networks with millions or billions of parameters which can then be trained with massive amounts of data. However, similar trend has largely eluded training of deep reinforcement learning (RL) algorithms where larger networks do not lead to performance improvement. Previous work has shown that this is mostly due to instability during training of deep RL agents when using larger networks. In this paper, we make an attempt to understand and address training of larger networks for deep RL. We first show that naively increasing network capacity does not improve performance. Then, we propose a novel method that consists of 1) wider networks with DenseNet connection, 2) decoupling representation learning from training of RL, 3) a distributed training method to mitigate overfitting problems. Using this three-fold technique, we show that we can train very large networks that result in significant performance gains. We present several ablation studies to demonstrate the efficacy of the proposed method and some intuitive understanding of the reasons for performance gain. We show that our proposed method outperforms other baseline algorithms on several challenging locomotion tasks.

Kei Ota, Devesh K. Jha, Tadashi Onishi et al.

The main novelty of the proposed approach is that it allows a robot to learn an end-to-end policy which can adapt to changes in the environment during execution. While goal conditioning of policies has been studied in the RL literature, such approaches are not easily extended to cases where the robot's goal can change during execution. This is something that humans are naturally able to do. However, it is difficult for robots to learn such reflexes (i.e., to naturally respond to dynamic environments), especially when the goal location is not explicitly provided to the robot, and instead needs to be perceived through a vision sensor. In the current work, we present a method that can achieve such behavior by combining traditional kinematic planning, deep learning, and deep reinforcement learning in a synergistic fashion to generalize to arbitrary environments. We demonstrate the proposed approach for several reaching and pick-and-place tasks in simulation, as well as on a real system of a 6-DoF industrial manipulator. A video describing our work could be found \url{https://youtu.be/hE-Ew59GRPQ}.

Ryo Yonetani, Tatsunori Taniai, Mohammadamin Barekatain et al.

We present Neural A*, a novel data-driven search method for path planning problems. Despite the recent increasing attention to data-driven path planning, machine learning approaches to search-based planning are still challenging due to the discrete nature of search algorithms. In this work, we reformulate a canonical A* search algorithm to be differentiable and couple it with a convolutional encoder to form an end-to-end trainable neural network planner. Neural A* solves a path planning problem by encoding a problem instance to a guidance map and then performing the differentiable A* search with the guidance map. By learning to match the search results with ground-truth paths provided by experts, Neural A* can produce a path consistent with the ground truth accurately and efficiently. Our extensive experiments confirmed that Neural A* outperformed state-of-the-art data-driven planners in terms of the search optimality and efficiency trade-off. Furthermore, Neural A* successfully predicted realistic human trajectories by directly performing search-based planning on natural image inputs. Project page: https://omron-sinicx.github.io/neural-astar/

Wonjik Kim, Asako Kanezaki, Masayuki Tanaka

The usage of convolutional neural networks (CNNs) for unsupervised image segmentation was investigated in this study. In the proposed approach, label prediction and network parameter learning are alternately iterated to meet the following criteria: (a) pixels of similar features should be assigned the same label, (b) spatially continuous pixels should be assigned the same label, and (c) the number of unique labels should be large. Although these criteria are incompatible, the proposed approach minimizes the combination of similarity loss and spatial continuity loss to find a plausible solution of label assignment that balances the aforementioned criteria well. The contributions of this study are four-fold. First, we propose a novel end-to-end network of unsupervised image segmentation that consists of normalization and an argmax function for differentiable clustering. Second, we introduce a spatial continuity loss function that mitigates the limitations of fixed segment boundaries possessed by previous work. Third, we present an extension of the proposed method for segmentation with scribbles as user input, which showed better accuracy than existing methods while maintaining efficiency. Finally, we introduce another extension of the proposed method: unseen image segmentation by using networks pre-trained with a few reference images without re-training the networks. The effectiveness of the proposed approach was examined on several benchmark datasets of image segmentation.

Kei Ota, Yoko Sasaki, Devesh K. Jha et al.

In this paper, we consider the problem of building learning agents that can efficiently learn to navigate in constrained environments. The main goal is to design agents that can efficiently learn to understand and generalize to different environments using high-dimensional inputs (a 2D map), while following feasible paths that avoid obstacles in obstacle-cluttered environment. To achieve this, we make use of traditional path planning algorithms, supervised learning, and reinforcement learning algorithms in a synergistic way. The key idea is to decouple the navigation problem into planning and control, the former of which is achieved by supervised learning whereas the latter is done by reinforcement learning. Specifically, we train a deep convolutional network that can predict collision-free paths based on a map of the environment-- this is then used by a reinforcement learning algorithm to learn to closely follow the path. This allows the trained agent to achieve good generalization while learning faster. We test our proposed method in the recently proposed Safety Gym suite that allows testing of safety-constraints during training of learning agents. We compare our proposed method with existing work and show that our method consistently improves the sample efficiency and generalization capability to novel environments.

Nevrez Imamoglu, Guanqun Ding, Yuming Fang et al.

Various saliency detection algorithms from color images have been proposed to mimic eye fixation or attentive object detection response of human observers for the same scenes. However, developments on hyperspectral imaging systems enable us to obtain redundant spectral information of the observed scenes from the reflected light source from objects. A few studies using low-level features on hyperspectral images demonstrated that salient object detection can be achieved. In this work, we proposed a salient object detection model on hyperspectral images by applying manifold ranking (MR) on self-supervised Convolutional Neural Network (CNN) features (high-level features) from unsupervised image segmentation task. Self-supervision of CNN continues until clustering loss or saliency maps converges to a defined error between each iteration. Finally, saliency estimations is done as the saliency map at last iteration when the self-supervision procedure terminates with convergence. Experimental evaluations demonstrated that proposed saliency detection algorithm on hyperspectral images is outperforming state-of-the-arts hyperspectral saliency models including the original MR based saliency model.

Nevrez Imamoglu, Wataru Shimoda, Chi Zhang et al.

Bottom-up and top-down visual cues are two types of information that helps the visual saliency models. These salient cues can be from spatial distributions of the features (space-based saliency) or contextual / task-dependent features (object based saliency). Saliency models generally incorporate salient cues either in bottom-up or top-down norm separately. In this work, we combine bottom-up and top-down cues from both space and object based salient features on RGB-D data. In addition, we also investigated the ability of various pre-trained convolutional neural networks for extracting top-down saliency on color images based on the object dependent feature activation. We demonstrate that combining salient features from color and dept through bottom-up and top-down methods gives significant improvement on the salient object detection with space based and object based salient cues. RGB-D saliency integration framework yields promising results compared with the several state-of-the-art-models.

Hirokatsu Kataoka, Soma Shirakabe, Yun He et al.

The paper gives futuristic challenges disscussed in the cvpaper.challenge. In 2015 and 2016, we thoroughly study 1,600+ papers in several conferences/journals such as CVPR/ICCV/ECCV/NIPS/PAMI/IJCV.

Katsunori Ohnishi, Atsushi Kanehira, Asako Kanezaki et al.

We present a novel dataset and a novel algorithm for recognizing activities of daily living (ADL) from a first-person wearable camera. Handled objects are crucially important for egocentric ADL recognition. For specific examination of objects related to users' actions separately from other objects in an environment, many previous works have addressed the detection of handled objects in images captured from head-mounted and chest-mounted cameras. Nevertheless, detecting handled objects is not always easy because they tend to appear small in images. They can be occluded by a user's body. As described herein, we mount a camera on a user's wrist. A wrist-mounted camera can capture handled objects at a large scale, and thus it enables us to skip object detection process. To compare a wrist-mounted camera and a head-mounted camera, we also develop a novel and publicly available dataset that includes videos and annotations of daily activities captured simultaneously by both cameras. Additionally, we propose a discriminative video representation that retains spatial and temporal information after encoding frame descriptors extracted by Convolutional Neural Networks (CNN).