ACE Robotics and KUAIKEDA Set Up Retail Robot Training Base for Convenience Stores
ACE Robotics and KUAIKEDA are training dozens of robots in sync on food preparation, packing, restocking and small-object handling, with plans for thousands of retail robots by 2027.

The Embodied Retail Skill Training Base, a joint project by ACE Robotics and KUAIKEDA, has been set up in Deqing, Zhejiang. Here, a large number of robots are being trained in completing a series of daily tasks at a typical convenience store. In terms of architecture, this is the first time that the dual-brain collaborative architecture and Embodied AI RSI (Recursive Self-Improvement) of ACE Robotics have been deployed in an instant retail setting. In order to enable the generation, optimization and sharing of retail skills that can be executed by a large number of robots, these skills are generated using a general model and the Kaiwu World Model, and then optimized and shared.
For the three types of skills, for food preparation and retrieval, robots learn to pick and serve macarons, pick and serve oden, and assemble and heat up to serve bun products in a microwave. Single tasks such as picking, processing and plating will be aggregated into a complete meal for hot food counters. For packing and store organization, robots practice mixed packing, sealing, labeling, checking for expiration dates, and restocking. The skills are aggregated into continuous workflows and are applied in the store.
The tens of retail skills in packing and in organizing on store shelves are combined into continuous workflows of retail skills, for example, of mixed packing, of sealing, of labeling, of checking expiration dates and of restocking on shelves in stores. The execution experience in all of these retail skills is transferred to store floors where the thousands of retail robots are working.
Small objects in tight spaces, such as transparent objects, objects with reflective surfaces, soft objects, etc. Tasks such as opening eyeglass cases, handling coffee and tea bags, stamping documents, sorting and arranging shopping baskets, etc. Combining these simple tasks into sorting, stamping and shelving tasks.
Tens of units of the robotics companies are learning to master all these tasks in sync in order to analyze how basic actions as well as the experience gained from executing them can be shared among robots of all units. The feedback from the actual machines in the store environment will be used for continuously improving the skills in the Training Base.
By 2027 ACE Robotics and KUAIKEDA plan to deploy in hundreds of stores thousands of retail robots which have been trained in Embodied Retail Skill Training Base in Deqing, Zhejiang. As the companies have not published yet any data on the success rates, task execution times, and number of stores where skills are transferred between retail robots and put into practice in working stores, it remains to be seen how effective such transfer is.
More articles on this topic
155 articles
ReportChinese Chip Manufacturer Expands Into Robotics and Physical AI
ReportFamilyMart to Roll Out Beverage-Restocking Robots Across Japan
ReportMegaRobo Raises $200 Million for AI-Powered Lab Automation
ReportHidonix Launches Hido II Robot for Indoor Mapping Trials
ReportZeroSquare Open-Sources Three Embodied AI Tools to Improve Robot Training with Real-World Data
ReportBrightAI Raises $97 Million to Expand AI Infrastructure Inspection
ReportRockwell Automation to Buy AGILOX, Expanding European Robot Business
ReportPostal Operators in Asia Expand Robotics for Parcel Sorting
ReportRainbow Robotics Partners With Samsung Unit on Factory Automation
ReportSPARKOZ Plans U.S. Cleaning Robot Production With Pacific Floorcare
ReportROBOTERA Deploys Humanoids at 11 China Post Hubs
ReportGeneral Motors Plans to Increase Factory Robotics and Automation by 20% by 2028BUSINESS NEWS WEEKLY LETTER
The Weekly Letter for Robotics Professionals, Summarizing the Most Important Industry Moves, Launches, Deals and Signals.