工业元宇宙场景下的人机智能体发展

随着国务院《新一代人工智能发展规划》的深入落实,以企业为主体的高质量AI场景创新,成为工业和信息化部等政策指导机构重点关注并大力支持的发展方向。2023年,大语言模型奠定了AI技术底座,进而拉低了基于自然语言理解的多模态数字化内容生成门槛,为元宇宙进化补上了数字化生产力升级的关键一环。AI的快速发展也引发了一系列伦理问题,甚至部分业界专家也频频警示“AI威胁论”。本文聚焦“工业元宇宙”场景,在重新定义物理“原宇宙”、数字“源宇宙”和认知“元宇宙”三元体系基础上,结合联想控股旗下企业及国外企业实践案例,探讨人机智能体的可控进化体系和面向意图的开发范式,为新型工业化的智能化升级和系统实施提供了可验证场景。

Three-dimensional multi-constraint route planning of unmanned aerial vehicle low-altitude penetration based on coevolutionary multi-agent genetic algorithm

To address the issue of premature convergence and slow convergence rate in three-dimensional (3D) route planning of unmanned aerial vehicle (UAV) low-altitude penetration,a novel route planning method was proposed.First and foremost,a coevolutionary multi-agent genetic algorithm (CE-MAGA) was formed by introducing coevolutionary mechanism to multi-agent genetic algorithm (MAGA),an efficient global optimization algorithm.A dynamic route representation form was also adopted to improve the flight route accuracy.Moreover,an efficient constraint handling method was used to simplify the treatment of multi-constraint and reduce the time-cost of planning computation.Simulation and corresponding analysis show that the planning results of CE-MAGA have better performance on terrain following,terrain avoidance,threat avoidance (TF/TA2) and lower route costs than other existing algorithms.In addition,feasible flight routes can be acquired within 2 s,and the convergence rate of the whole evolutionary process is very fast.

Multi-agent reinforcement learning with cooperation based on eligibility traces

The application of reinforcement learning is widely used by multi-agent systems in recent years. An agent uses a multi-agent system to cooperate with other agents to accomplish the given task, and one agent′s behavior usually affects the others′ behaviors. In traditional reinforcement learning, one agent takes the others location, so it is difficult to consider the others′ behavior, which decreases the learning efficiency. This paper proposes multi-agent reinforcement learning with cooperation based on eligibility traces, i.e. one agent estimates the other agent′s behavior with the other agent′s eligibility traces. The results of this simulation prove the validity of the proposed learning method.

Organic crystal-based flexible smart materials

Although the famous brittle characteristics of molecular crystals are unfavorable when they are used as flexible smart materials(FSMs),an increasing number of organic crystal-based FSMs have been reported recently.This breaks the perception of their stiff and brittle properties and promises a bright future for basic research and practical applications.Crystalline smart materials present considerable advantages over polymer materials under certain circumstances,rendering them potential candidates for certain applications,such as rapidly responsive actuators,ON/OFF switching,and microrobots.In this review,we summarize the recent developments in the field of organic crystal-based FSMs,including the derivatives of azobenzene,diarylethene,anthracene,and olefin.These organic crystal-based FSMs can bend,curl,twist,deform,or respond otherwise to external stimuli,such as heat or light.The detailed mechanisms of their smart behaviors are discussed with their potential applications in exciting intelligent fields.We believe this review could provide guidelines toward future fabrication and developments for novel organic crystal-based FSMs and their advanced smart applications.

New advances in complex motion control for single robot systems and multi-agent systems

CHEN Jie,email：chenjie@bit.edu.cn

A fast face detection architecture for auto-focus in smart-phones and digital cameras

Auto-focus is very important for capturing sharp human face centered images in digital and smart phone cameras. With the development of image sensor technology, these cameras support more and more highresolution images to be processed. Currently it is difficult to support fast auto-focus at low power consumption on high-resolution images. This work proposes an efficient architecture for an Ada Boost-based face-priority auto-focus. The architecture supports block-based integral image computation to improve the processing speed on high-resolution images; meanwhile, it is reconfigurable so that it enables the sub-window adaptive cascade classification, which greatly improves the processing speed and reduces power consumption. Experimental results show that 96% detection rate in average and 58 fps(frame per second) detection speed are achieved for the1080p(1920×1080) images. Compared with the state-of-the-art work, the detection speed is greatly improved and power consumption is largely reduced.