Completing complex tasks in unpredictable settings challenges robotic systems, requiring a step change in machine intelligence. Sensorimotor abilities are considered integral to human intelligence. Thus, biologically inspired machine intelligence might usefully combine artificial intelligence with robotic sensorimotor capabilities. Here we report an embodied large-language-model-enabled robot (ELLMER) framework, utilizing GPT-4 and a retrieval-augmented generation infrastructure, to enable robots to complete long-horizon tasks in unpredictable settings. The method extracts contextually relevant examples from a knowledge base, producing action plans that incorporate force and visual feedback and enabling adaptation to changing conditions. We tested ELLMER on a robot tasked with coffee making and plate decoration; these tasks consist of a sequence of sub-tasks from drawer opening to pouring, each benefiting from distinct feedback types and methods. We show that the ELLMER framework allows the robot to complete the tasks. This demonstration marks progress towards scalable, efficient and ‘intelligent robots’ able to complete complex tasks in uncertain environments.

Semantic segmentation is a process of linking each pixel in an image to a class label, and is widely used in the field of autonomous vehicles and robotics. Although deep learning methods have already made great progress for semantic segmentation, they either achieve great results with numerous parameters or design lightweight models but heavily sacrifice the segmentation accuracy. Because of the strict requirements of real-world applications, it is critical to design an effective real-time model with both competitive segmentation accuracy and small model capacity. In this paper, we propose a lightweight network named DABNet, which employs Depth-wise Asymmetric Bottleneck (DAB) and Point-wise Aggregation Decoder (PAD) module to tackle the challenging real-time semantic segmentation in urban scenes. Specifically, the DAB module creates a sufficient receptive field and densely utilizes the contextual information, and the PAD module aggregates the feature maps of different scales to optimize performance through the attention mechanism. Compared with existing methods, our network substantially reduces the number of parameters but still achieves high accuracy with real-time inference ability. Extensive ablation experiments on two challenging urban scene datasets (Cityscapes and CamVid) have proved the effectiveness of the proposed approach in real-time semantic segmentation.

World models, which are predictive representations of how environments evolve under actions, have become a central component of robot learning. They support policy learning, planning, simulation, evaluation, data generation, and have advanced rapidly with the rise of foundation models and large-scale video generation. However, the literature remains fragmented across architectures, functional roles, and embodied application domains. To address this gap, we present a comprehensive review of world models from a robot-learning perspective. We examine how world models are coupled with robot policies, how they serve as learned simulators for reinforcement learning and evaluation, and how robotic video world models have progressed from imagination-based generation to controllable, structured, and foundation-scale formulations. We further summarize representative datasets, benchmarks, and evaluation protocols. Overall, this survey systematically reviews the rapidly growing literature on world models for robot learning, clarifies key paradigms and applications, and highlights major challenges and future directions for predictive modeling in embodied agents. To facilitate continued access to newly emerging works, benchmarks, and resources, we will maintain and regularly update the accompanying GitHub repository on https://github.com/NTUMARS/Awesome-World-Model-for-Robotics-Policy alongside this survey.

With increasing attention to sustainable material development, bio-based materials such as bacterial cellulose (BC) have been widely explored in design research. However, their broader application remains constrained by long production cycles and limited controllability. To address these limitations, this study presents an exploratory investigation of AI-based virtual biomaterial generation under limited-data conditions. Specifically, an image dataset was first constructed from self-cultivated BC samples produced under controlled settings. Three generative adversarial network (GAN)-based models were trained and evaluated for virtual material generation. User perception experiments revealed that StyleGAN-generated samples received the highest ratings in naturalness, sense of growth, and credibility, approaching evaluations of physical BC samples. In addition, an image-to-3D proof-of-concept showed that GAN-generated textures could be used as visual inputs for preliminary 3D shape generation, while cross-material experiments demonstrated the potential for simulating novel material combinations. The findings indicate that, despite limited training data, GAN-generated textiles can capture key visual characteristics of BC and facilitate early-stage design exploration. By integrating biological experimentation with computational generation, this work proposes a AI-assisted framework with the potential to reduce physical material consumption and support sustainable design innovation. The study further suggests that generative models can function not only as visualization tools but also as assistive systems for future material design.