Visual Inspection
极智·视觉魔盒SmartVision Magic Box Pro
看见现场异常:缺陷、异物、人员行为、设备状态和安全风险。Sees site anomalies: defects, foreign objects, personnel behavior, equipment states, and safety risks.
AI Agent Native Products
围绕工业现场智能服务,StarAgent 形成视觉检测与预测性维护两条产品方向。平板端 AI Agent 负责自然语言交互与任务编排,嵌入式 AI 网关负责现场数据采集、边缘推理、报警联动和记录留存。 For industrial-site smart services, StarAgent covers two product directions: visual inspection and predictive maintenance. The tablet AI agent handles natural-language interaction and orchestration, while the embedded AI gateway handles field data, edge inference, alert linkage, and records.
Product Matrix
StarAgent 不是把参数堆给操作工,而是把视觉、传感器、边缘推理和工单流程组织成能对话的现场服务。StarAgent does not expose a pile of parameters to operators. It organizes vision, sensors, edge inference, and work-order flows into a conversational site service.
一个负责“看现场”,一个负责“懂设备”。One sees the site, the other understands equipment health.
Visual Inspection
看见现场异常:缺陷、异物、人员行为、设备状态和安全风险。Sees site anomalies: defects, foreign objects, personnel behavior, equipment states, and safety risks.
Predictive Maintenance
理解设备趋势:振动、温度、电流、压力、健康评分和停机风险。Understands equipment trends: vibration, temperature, current, pressure, health scores, and downtime risks.
四个问题对应现场落地闭环,不是四个独立产品。These four questions are deployment scenarios, not separate products.
场景 01Scenario 01
相机采集现场画面,边缘模型标出异常位置并保存样图。Cameras capture the scene; edge models mark anomaly locations and save sample images.
场景 02Scenario 02
融合振动、温度、电流等数据,给出健康趋势和维护窗口。Combines vibration, temperature, and current data into health trends and maintenance windows.
场景 03Scenario 03
Agent 用自然语言解释原因,给出巡检步骤、复核建议和工单内容。The agent explains causes, suggests inspection steps, and drafts work-order content.
场景 04Scenario 04
按规则触发报警、停线建议、现场控制或工单流转,记录全程留痕。Rules trigger alerts, stop-line suggestions, field control, or work-order flows with traceable records.