智慧餐桌清理機器人 Intelligent Table Clearing Robot 2026/07/14

簡介 Introductions

工研院與凌群電腦共同研發餐桌清理機器人，搭載端到端 VLA（視覺-語言-動作）模型，成功突破現有服務機器人「僅能送餐、無法清理」的技術瓶頸，有效緩解餐飲外場的人力短缺問題。該系統具備自動化桌面清理能力，能透過多模態感知識別餐桌上的餐具配置，並動態生成夾取與回收的動作路徑；在多變的餐飲環境中，能兼顧效率與穩定度，大幅提升清潔流程的一致性。

ITRI and The SYSCOM Group have jointly developed a table-clearing robot equipped with an end-to-end VLA (Vision-Language-Action) model. This technology breaks through the bottleneck of current service robots that are limited to food delivery, effectively alleviating front-of-house labor shortages. The system features automated table-clearing capabilities, using multimodal perception to identify tableware arrangements and dynamically generate action paths for grasping and clearing, balancing efficiency and stability in dynamic environments, , thereby significantly improving the consistency of the cleaning process.

特色與創新 Characters and Innovations

• 高適應性動態場景控制：透過真人示範資料集進行後訓練，在隨機動態環境中展現優異的適應力與泛化能力，超越傳統控制方法。
• AI 小腦 (VLA 視覺語言行為模型)：根據使用者下達的任務指令，結合實際場景，即時驅動機械手臂執行迅速、精準且穩定的清理動作。

• High-Adaptability Dynamic Control: Trained on real human data, excels in adaptability and generalization, outperforming traditional methods in dynamic environments.
• AI Cerebellum (VLA): Converts high-level tasks into precise real-time actions, enabling fast, accurate, and stable table clearing.

應用與效益 Applications and Benefits

• 提供送餐與用膳後桌面清理功能，全面實現外場自動化服務。
• 服務效率提升 30-40%，人力成本降低 60-70%。

• Enables cuisine delivery and full table clearing afterward, achieving complete front-of-house automation.
• Boosts service efficiency by 30-40% and reduces labor costs by 60-70%.