Multimodal Perception and Real-Time Fusion of Biomimetic Robots

Perception in intelligent robots is shifting away from single-modality sensing toward the coordinated use of multiple information channels. Focusing on intelligent bionic robots, this paper reviews how bionics and embodied intelligence have shaped the joint design of sensing hardware and perceptual algorithms. Then further investigates three representative paradigms for multimodal information fusion, namely point-cloud fusion, heterogeneous fusion, and bird’s-eye-view (BEV) fusion. In response to current limitations, especially insufficient device-level integration and weak cross-scene generalization, this paper provided a future development route centered on integrated hardware that combines communication, sensing, and transmission, together with optoelectronic chips and generative artificial intelligence models, in order to improve overall system performance and adaptive capability. The study is intended to provide a systematic reference for the advancement of multimodal perception in intelligent bionic robots and to support further progress toward stronger environmental understanding and more capable autonomous decision-making.