Object manipulation automation in logistic warehouses has recently been actively researched. However, shelf replenishment is a challenge that requires the precise and careful handling of densely piled objects. The irregular arrangement of objects on a shelf makes this task particularly difficult. This paper presents an approach for generating a safe replenishment process from a single depth image, which is provided as an input to two networks to identify arrangement patterns and predict the occurrence of collapsing objects. The proposed inference-based strategy provides an appropriate decision and course of action on whether to create an insertion space while considering the safety of the shelf content. In particular, we exploit the bimanual dexterous manipulation capabilities of the associated robot to resolve the task safely, without re-organizing the entire shelf. Experiments with a real bimanual robot were performed in three typical scenarios: shelved, stacked, and random. The objects were randomly placed in each scenario. The experimental results verify the performance of our proposed method in randomized situations on a shelf with a real bimanual robot.

To automate operations in a logistic warehouse, a robot needs to extract items from the clutter on a shelf without collapsing the clutter. To address this problem, this study proposes a multi-step motion planner to stably extract an item by using the support relations of each object included in the clutter. This study primarily focuses on safe extraction, which allows the robot to choose the best next action based on limited observations. By estimating the support relations, we construct a collapse prediction graph to obtain the appropriate order of object extraction. Thus, the target object can be extracted without collapsing the pile. Furthermore, we show that the efficiency of the robot is improved if it uses one of its arms to extract the target object while the other supports a neighboring object. The proposed method is evaluated in real-world experiments on detecting support relations and object extraction tasks. This study makes a significant contribution because the experimental results indicate that the robot can estimate support relations based on collapse predictions and perform safe extraction in real environments. Our multi-step extraction plan ensures both better performance and robustness to achieve safe object extraction tasks from the clutter.

The development of large language models and vision-language models (VLMs) has resulted in the increasing use of robotic systems in various fields. However, the effective integration of these models into real-world robotic tasks is a key challenge. We developed a versatile robotic system called SuctionPrompt that utilizes prompting techniques of VLMs combined with 3D detections to perform product-picking tasks in diverse and dynamic environments. Our method highlights the importance of integrating 3D spatial information with adaptive action planning to enable robots to approach and manipulate objects in novel environments. In the validation experiments, the system accurately selected suction points 75.4%, and achieved a 65.0% success rate in picking common items. This study highlights the effectiveness of VLMs in robotic manipulation tasks, even with simple 3D processing.

This survey provides an in-depth analysis of recent advancements in foundational models that leverage large and diverse datasets to enable flexible and general-purpose manipulation skills. We reviewed the main concepts of Vision-Language-Action (VLA) models with emphasis on their connection to broader trends in robot learning and the architectural strategies that facilitate integration across vision, language, and action. The survey contains current training methodologies, dataset construction techniques, and the role of high-quality demonstrations in achieving reliable performance across diverse tasks. Key trends are identified, including the adoption of open-source tools, scaling of model size and training data, and improved generalization across platforms. In addition, we noted that VLA models demonstrate strong performance in controlled environments; however, significant challenges persist in long-term task planning, failure recovery, and operation in complex real-world scenarios. This survey aims to serve as a comprehensive reference for researchers and outline future research directions that may advance the practical deployment of VLA models in robotic manipulation.

Robots that can operate autonomously in a human living environment are necessary to have the ability to handle various tasks flexibly. One crucial element is coordinated bimanual movements that enable functions that are difficult to perform with one hand alone. In recent years, learning-based models that focus on the possibilities of bimanual movements have been proposed. However, the high degree of freedom of the robot makes it challenging to reason about control, and the left and right robot arms need to adjust their actions depending on the situation, making it difficult to realize more dexterous tasks. To address the issue, we focus on coordination and efficiency between both arms, particularly for synchronized actions. Therefore, we propose a novel imitation learning architecture that predicts cooperative actions. We differentiate the architecture for both arms and add an intermediate encoder layer, Inter-Arm Coordinated transformer Encoder (IACE), that facilitates synchronization and temporal alignment to ensure smooth and coordinated actions. To verify the effectiveness of our architectures, we perform distinctive bimanual tasks. The experimental results showed that our model demonstrated a high success rate for comparison and suggested a suitable architecture for the policy learning of bimanual manipulation.