Automatically extracting chemical entities from unstructured literature and patents, chemical named entity recognition (ChemNER) provides the foundational data required for constructing chemical knowledge graphs. As chemical entities frequently contain special characters and complex nested structures, existing methods often overlook their distinctive character-level distribution patterns. This omission poses substantial challenges for conventional approaches, including sequence labeling, span classification, and large language models (LLMs), in accurately identifying such intricate entities. This paper introduces CORE-NER (Character-Oriented Reference-Enhanced NER), a novel framework for chemical named entity recognition. The method begins by pre-extracting candidate chemical entities from the input text and retrieving semantically similar reference entities from a chemical entity knowledge base. We further propose a character-level masking strategy that explicitly captures the character distribution patterns and local dependency features of reference entities via a masked-prediction mechanism, thereby enabling learning of fine-grained character-level contextual representations. By integrating the resulting character-aware features with the original textual representations, the model incorporates explicit knowledge of the internal specialized structures of chemical entities without introducing significant computational overhead. The study employs the LoRA (Low-Rank Adaptation) parameter-efficient fine-tuning framework combined with task-specific instruction optimization to enhance the generative performance of large language models in specialized domains. Experimental results show that our approach consistently outperforms traditional baselines and existing LLM solutions across four benchmark chemical datasets, confirming the effectiveness and practical value of the proposed method. Our code is available for access at https://github.com/YanDDDeat/CORENER.

This paper proposes a configurable and scalable framework based on Controlled Robot Language with Frame Semantics (FrameCRL) for plan generation. Given natural language instructions, FrameCRL constructs an equivalent formal semantic formulation in the form of discourse representation structures (DRS). Imperative verbs are extracted from the semantic structures as keys to anchor relevant semantic frames from FrameNet, and the selected semantic frames are used to construct goal statements in planning language. Non-imperative statements are further analyzed to generate object specifications and the initial state of the planning problem. These generated statements are then merged into a single planning script, which can be solved directly by the integrated planner. The performance of FrameCRL was evaluated on various natural language corpora and compared with large language models (LLM) based methods in plan generation. The results demonstrated the outperformance of FrameCRL in generating high-quality plans and its capability to handle large context scenarios. The FrameCRL was also tested on pick-and-place tasks using a dual-arm robot and it showcased a robust performance in linguistic understanding.

Autonomous industrial assembly relies on the precise measurement of spatial constraints as designed by computer-aided design (CAD) software such as SolidWorks. This paper proposes a framework for an intelligent industrial robot to understand the spatial constraints for model assembly. An extended generative adversary network (GAN) with a 3D long short-term memory (LSTM) network was designed to composite 3D point clouds from a single RGB-D scan. The spatial constraints of the segmented point clouds are identified by a neural-logic network that incorporates general knowledge of spatial constraints in terms of first-order logic. The model was designed to comprehend a complete set of spatial constraints that are consistent with industrial CAD software, including left, right, above, below, front, behind, parallel, perpendicular, concentric, and coincident relations. The accuracy of 3D model composition and spatial constraint identification was evaluated by the RGB-D scans and 3D models in the ABC dataset. The proposed model achieved 57.23% intersection over union (IoU) in 3D model composition, and over 99% in comprehending all spatial constraints.

Understanding spatial relations of objects is critical in many robotic applications such as grasping, manipulation, and obstacle avoidance. Humans can simply reason object's spatial relations from a glimpse of a scene based on prior knowledge of spatial constraints. The proposed method enables a robot to comprehend spatial relationships among objects from RGB-D data. This paper proposed a neural-logic learning framework to learn and reason spatial relations from raw data by following logic rules on spatial constraints. The neural-logic network consists of three blocks: grounding block, spatial logic block, and inference block. The grounding block extracts high-level features from the raw sensory data. The spatial logic blocks can predicate fundamental spatial relations by training a neural network with spatial constraints. The inference block can infer complex spatial relations based on the predicated fundamental spatial relations. Simulations and robotic experiments evaluated the performance of the proposed method.