Atsushi Kanehira

Naoki Wake, Atsushi Kanehira, Kazuhiro Sasabuchi et al.

We introduce a pipeline that enhances a general-purpose Vision Language Model, GPT-4V(ision), to facilitate one-shot visual teaching for robotic manipulation. This system analyzes videos of humans performing tasks and outputs executable robot programs that incorporate insights into affordances. The process begins with GPT-4V analyzing the videos to obtain textual explanations of environmental and action details. A GPT-4-based task planner then encodes these details into a symbolic task plan. Subsequently, vision systems spatially and temporally ground the task plan in the videos. Objects are identified using an open-vocabulary object detector, and hand-object interactions are analyzed to pinpoint moments of grasping and releasing. This spatiotemporal grounding allows for the gathering of affordance information (e.g., grasp types, waypoints, and body postures) critical for robot execution. Experiments across various scenarios demonstrate the method's efficacy in enabling real robots to operate from one-shot human demonstrations. Meanwhile, quantitative tests have revealed instances of hallucination in GPT-4V, highlighting the importance of incorporating human supervision within the pipeline. The prompts of GPT-4V/GPT-4 are available at this project page: https://microsoft.github.io/GPT4Vision-Robot-Manipulation-Prompts/

Naoki Wake, Atsushi Kanehira, Kazuhiro Sasabuchi et al.

This technical report explores the ability of ChatGPT in recognizing emotions from text, which can be the basis of various applications like interactive chatbots, data annotation, and mental health analysis. While prior research has shown ChatGPT's basic ability in sentiment analysis, its performance in more nuanced emotion recognition is not yet explored. Here, we conducted experiments to evaluate its performance of emotion recognition across different datasets and emotion labels. Our findings indicate a reasonable level of reproducibility in its performance, with noticeable improvement through fine-tuning. However, the performance varies with different emotion labels and datasets, highlighting an inherent instability and possible bias. The choice of dataset and emotion labels significantly impacts ChatGPT's emotion recognition performance. This paper sheds light on the importance of dataset and label selection, and the potential of fine-tuning in enhancing ChatGPT's emotion recognition capabilities, providing a groundwork for better integration of emotion analysis in applications using ChatGPT.

Naoki Wake, Atsushi Kanehira, Kazuhiro Sasabuchi et al.

Video action localization aims to find the timings of specific actions from a long video. Although existing learning-based approaches have been successful, they require annotating videos, which comes with a considerable labor cost. This paper proposes a training-free, open-vocabulary approach based on emerging off-the-shelf vision-language models (VLMs). The challenge stems from the fact that VLMs are neither designed to process long videos nor tailored for finding actions. We overcome these problems by extending an iterative visual prompting technique. Specifically, we sample video frames and create a concatenated image with frame index labels, allowing a VLM to identify the frames that most likely correspond to the start and end of the action. By iteratively narrowing the sampling window around the selected frames, the estimation gradually converges to more precise temporal boundaries. We demonstrate that this technique yields reasonable performance, achieving results comparable to state-of-the-art zero-shot action localization. These results support the use of VLMs as a practical tool for understanding videos. Sample code is available at https://microsoft.github.io/VLM-Video-Action-Localization/

Naoki Wake, Atsushi Kanehira, Kazuhiro Sasabuchi et al.

This technical paper introduces a chatting robot system that utilizes recent advancements in large-scale language models (LLMs) such as GPT-3 and ChatGPT. The system is integrated with a co-speech gesture generation system, which selects appropriate gestures based on the conceptual meaning of speech. Our motivation is to explore ways of utilizing the recent progress in LLMs for practical robotic applications, which benefits the development of both chatbots and LLMs. Specifically, it enables the development of highly responsive chatbot systems by leveraging LLMs and adds visual effects to the user interface of LLMs as an additional value. The source code for the system is available on GitHub for our in-house robot (https://github.com/microsoft/LabanotationSuite/tree/master/MSRAbotChatSimulation) and GitHub for Toyota HSR (https://github.com/microsoft/GPT-Enabled-HSR-CoSpeechGestures).

Naoki Wake, Atsushi Kanehira, Jun Takamatsu et al.

The use of Large Language Models (LLMs) for generating Behavior Trees (BTs) has recently gained attention in the robotics community, yet remains in its early stages of development. In this paper, we propose a novel framework that leverages Vision-Language Models (VLMs) to interactively generate and edit BTs that address visual conditions, enabling context-aware robot operations in visually complex environments. A key feature of our approach lies in the conditional control through self-prompted visual conditions. Specifically, the VLM generates BTs with visual condition nodes, where conditions are expressed as free-form text. Another VLM process integrates the text into its prompt and evaluates the conditions against real-world images during robot execution. We validated our framework in a real-world cafe scenario, demonstrating both its feasibility and limitations.

Kazuhiro Sasabuchi, Naoki Wake, Atsushi Kanehira et al.

Recent large language models (LLMs) are capable of planning robot actions. In this paper, we explore how LLMs can be used for planning actions with tasks involving situational human-robot interaction (HRI). A key problem of applying LLMs in situational HRI is balancing between "respecting the current human's activity" and "prioritizing the robot's task," as well as understanding the timing of when to use the LLM to generate an action plan. In this paper, we propose a necessary plan-and-act skill design to solve the above problems. We show that a critical factor for enabling a robot to switch between passive / active interaction behavior is to provide the LLM with an action text about the current robot's action. We also show that a second-stage question to the LLM (about the next timing to call the LLM) is necessary for planning actions at an appropriate timing. The skill design is applied to an Engage skill and is tested on four distinct interaction scenarios. We show that by using the skill design, LLMs can be leveraged to easily scale to different HRI scenarios with a reasonable success rate reaching 90% on the test scenarios.

Naoki Wake, Atsushi Kanehira, Daichi Saito et al.

Multi-step dexterous manipulation is a fundamental skill in household scenarios, yet remains an underexplored area in robotics. This paper proposes a modular approach, where each step of the manipulation process is addressed with dedicated policies based on effective modality input, rather than relying on a single end-to-end model. To demonstrate this, a dexterous robotic hand performs a manipulation task involving picking up and rotating a box. Guided by insights from neuroscience, the task is decomposed into three sub-skills, 1)reaching, 2)grasping and lifting, and 3)in-hand rotation, based on the dominant sensory modalities employed in the human brain. Each sub-skill is addressed using distinct methods from a practical perspective: a classical controller, a Vision-Language-Action model, and a reinforcement learning policy with force feedback, respectively. We tested the pipeline on a real robot to demonstrate the feasibility of our approach. The key contribution of this study lies in presenting a neuroscience-inspired, modality-driven methodology for multi-step dexterous manipulation.

Kazuhiro Sasabuchi, Naoki Wake, Atsushi Kanehira et al.

In human-robot interaction (HRI), the beginning of an interaction is often complex. Whether the robot should communicate with the human is dependent on several situational factors (e.g., the current human's activity, urgency of the interaction, etc.). We test whether large language models (LLM) and vision language models (VLM) can provide solutions to this problem. We compare four different system-design patterns using LLMs and VLMs, and test on a test set containing 84 human-robot situations. The test set mixes several publicly available datasets and also includes situations where the appropriate action to take is open-ended. Our results using the GPT-4o and Phi-3 Vision model indicate that LLMs and VLMs are capable of handling interaction beginnings when the desired actions are clear, however, challenge remains in the open-ended situations where the model must balance between the human and robot situation.

Jun Takamatsu, Atsushi Kanehira, Kazuhiro Sasabuchi et al.

Robots are strongly expected as a means of replacing human tasks. If a robot has a human-like physicality, the possibility of replacing human tasks increases. In the case of household service robots, it is desirable for them to be on a human-like size so that they do not become excessively large in order to coexist with humans in their operating environment. However, robots with size limitations tend to have difficulty solving inverse kinematics (IK) due to mechanical limitations, such as joint angle limitations. Conversely, if the difficulty coming from this limitation could be mitigated, one can expect that the use of such robots becomes more valuable. In numerical IK solver, which is commonly used for robots with higher degrees-of-freedom (DOF), the solvability of IK depends on the initial guess given to the solver. Thus, this paper proposes a method for generating a good initial guess for a numerical IK solver given the target hand configuration. For the purpose, we define the goodness of an initial guess using the scaled Jacobian matrix, which can calculate the manipulability index considering the joint limits. These two factors are related to the difficulty of solving IK. We generate the initial guess by optimizing the goodness using the genetic algorithm (GA). To enumerate much possible IK solutions, we use the reachability map that represents the reachable area of the robot hand in the arm-base coordinate system. We conduct quantitative evaluation and prove that using an initial guess that is judged to be better using the goodness value increases the probability that IK is solved. Finally, as an application of the proposed method, we show that by generating good initial guesses for IK a robot actually achieves three typical scenarios.

Naoki Wake, Atsushi Kanehira, Kazuhiro Sasabuchi et al.

Self-handover, transferring an object between one's own hands, is a common but understudied bimanual action. While it facilitates seamless transitions in complex tasks, the strategies underlying its execution remain largely unexplored. Here, we introduce the first systematic taxonomy of self-handover, derived from manual annotation of over 12 hours of cooking activity performed by 21 participants. Our analysis reveals that self-handover is not merely a passive transition, but a highly coordinated action involving anticipatory adjustments by both hands. As a step toward automated analysis of human manipulation, we further demonstrate the feasibility of classifying self-handover types using a state-of-the-art vision-language model. These findings offer fresh insights into bimanual coordination, underscoring the role of self-handover in enabling smooth task transitions-an ability essential for adaptive dual-arm robotics.

Tommi Kerola, Jie Li, Atsushi Kanehira et al.

Panoptic segmentation brings together two separate tasks: instance and semantic segmentation. Although they are related, unifying them faces an apparent paradox: how to learn simultaneously instance-specific and category-specific (i.e. instance-agnostic) representations jointly. Hence, state-of-the-art panoptic segmentation methods use complex models with a distinct stream for each task. In contrast, we propose Hierarchical Lovász Embeddings, per pixel feature vectors that simultaneously encode instance- and category-level discriminative information. We use a hierarchical Lovász hinge loss to learn a low-dimensional embedding space structured into a unified semantic and instance hierarchy without requiring separate network branches or object proposals. Besides modeling instances precisely in a proposal-free manner, our Hierarchical Lovász Embeddings generalize to categories by using a simple Nearest-Class-Mean classifier, including for non-instance "stuff" classes where instance segmentation methods are not applicable. Our simple model achieves state-of-the-art results compared to existing proposal-free panoptic segmentation methods on Cityscapes, COCO, and Mapillary Vistas. Furthermore, our model demonstrates temporal stability between video frames.

Atsushi Kanehira, Tatsuya Harada

This paper addresses the generation of explanations with visual examples. Given an input sample, we build a system that not only classifies it to a specific category, but also outputs linguistic explanations and a set of visual examples that render the decision interpretable. Focusing especially on the complementarity of the multimodal information, i.e., linguistic and visual examples, we attempt to achieve it by maximizing the interaction information, which provides a natural definition of complementarity from an information theoretical viewpoint. We propose a novel framework to generate complemental explanations, on which the joint distribution of the variables to explain, and those to be explained is parameterized by three different neural networks: predictor, linguistic explainer, and example selector. Explanation models are trained collaboratively to maximize the interaction information to ensure the generated explanation are complemental to each other for the target. The results of experiments conducted on several datasets demonstrate the effectiveness of the proposed method.

Atsushi Kanehira, Kentaro Takemoto, Sho Inayoshi et al.

This study addresses generating counterfactual explanations with multimodal information. Our goal is not only to classify a video into a specific category, but also to provide explanations on why it is not categorized to a specific class with combinations of visual-linguistic information. Requirements that the expected output should satisfy are referred to as counterfactuality in this paper: (1) Compatibility of visual-linguistic explanations, and (2) Positiveness/negativeness for the specific positive/negative class. Exploiting a spatio-temporal region (tube) and an attribute as visual and linguistic explanations respectively, the explanation model is trained to predict the counterfactuality for possible combinations of multimodal information in a post-hoc manner. The optimization problem, which appears during training/inference, can be efficiently solved by inserting a novel neural network layer, namely the maximum subpath layer. We demonstrated the effectiveness of this method by comparison with a baseline of the action recognition datasets extended for this task. Moreover, we provide information-theoretical insight into the proposed method.

Atsushi Kanehira, Luc Van Gool, Yoshitaka Ushiku et al.

This paper introduces a novel variant of video summarization, namely building a summary that depends on the particular aspect of a video the viewer focuses on. We refer to this as $\textit{viewpoint}$. To infer what the desired $\textit{viewpoint}$ may be, we assume that several other videos are available, especially groups of videos, e.g., as folders on a person's phone or laptop. The semantic similarity between videos in a group vs. the dissimilarity between groups is used to produce $\textit{viewpoint}$-specific summaries. For considering similarity as well as avoiding redundancy, output summary should be (A) diverse, (B) representative of videos in the same group, and (C) discriminative against videos in the different groups. To satisfy these requirements (A)-(C) simultaneously, we proposed a novel video summarization method from multiple groups of videos. Inspired by Fisher's discriminant criteria, it selects summary by optimizing the combination of three terms (a) inner-summary, (b) inner-group, and (c) between-group variances defined on the feature representation of summary, which can simply represent (A)-(C). Moreover, we developed a novel dataset to investigate how well the generated summary reflects the underlying $\textit{viewpoint}$. Quantitative and qualitative experiments conducted on the dataset demonstrate the effectiveness of proposed method.

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).

Ken Miura, Tetsuaki Mano, Atsushi Kanehira et al.

MILJS is a collection of state-of-the-art, platform-independent, scalable, fast JavaScript libraries for matrix calculation and machine learning. Our core library offering a matrix calculation is called Sushi, which exhibits far better performance than any other leading machine learning libraries written in JavaScript. Especially, our matrix multiplication is 177 times faster than the fastest JavaScript benchmark. Based on Sushi, a machine learning library called Tempura is provided, which supports various algorithms widely used in machine learning research. We also provide Soba as a visualization library. The implementations of our libraries are clearly written, properly documented and thus can are easy to get started with, as long as there is a web browser. These libraries are available from http://mil-tokyo.github.io/ under the MIT license.