---
id: "lib-2026-10-28-sichuan-s-top-convenience-chain-tried-robot-sales"
type: "library"
title: "Sichuan's top convenience chain tried robot sales, and it failed"
title_original: "四川最牛便利连锁超市搞机器人售货，黄了"
lang: "ZH"
org: "量子位 QbitAI"
source_type: "Practitioner post"
date: "2026-10-28"
week: "2026-W44"
url: "https://nosov.ai/library/lib-2026-10-28-sichuan-s-top-convenience-chain-tried-robot-sales/"
original_url: "https://www.qbitai.com/2026/10/505955.html"
tier: "B"
tier_name: "Useful, limited"
tier_reason: "Original QbitAI ROBO reporting that combines a listed retailer's public filings with site visits and industry interviews on robot retail; useful on-the-ground detail, but most figures come from unnamed or single-name sources and media reports."
summary_basis: "full-text"
original_minutes: "13"
tags: ["business transformation", "retail robotics", "deployment failure"]
---

# Sichuan's top convenience chain tried robot sales, and it failed

Original title: 四川最牛便利连锁超市搞机器人售货，黄了

Tier B (Useful, limited): Original QbitAI ROBO reporting that combines a listed retailer's public filings with site visits and industry interviews on robot retail; useful on-the-ground detail, but most figures come from unnamed or single-name sources and media reports.

Summary based on the full text.

## In short

A QbitAI ROBO feature (28 October 2026) on Hongqi Chain, Sichuan's leading listed convenience-store chain, which announced a robot unattended-sales system in March 2025 and ended it after 16 months. The robots could not reliably grasp mixed items on dense shelves and dropped wet glass bottles of cold drinks. Hongqi switched to scaling a 24-hour remote ("cloud") staffing model. Visits to robot kiosks in Beijing show robots acting mainly as slow pick-up arms with a sales patter, staffed by people throughout opening hours, with tourist sites as the main buyers.

## Key numbers

- about 35.0%: rise in Hongqi Chain's share price from the March 2025 robot announcement to the 23 April close
- 16 months: time from project start to termination, announced in early July 2026
- about 2.87 million yuan: Hongqi's 2025 R&D expense (about 2.87 million yuan), which the reporters infer is roughly the project's spend
- nearly 700,000 yuan: e-commerce listing price of a retail robot model in use; a full kiosk is estimated near 1 million yuan
- 60-70%: share of robot retail kiosks placed at tourist sites and city landmarks, per a former UBTech sales manager

## What it reports

- Hongqi's risk notice listed three gaps: high grasp failure on dense shelves and odd-shaped goods; vision errors under varied lighting; the combined demands of obstacle avoidance and stable data links. Lab tests were complete, but "the lab and real use differ greatly".
- July 2026 termination notice: the system had not reached large-scale commercial standards; the decision cited return on investment. A source close to the firm said the aim was night-time staff replacement.
- Hongqi did not disclose project cost; its 2025 report shows R&D expense of about 2.87 million yuan, and earlier reports showed none, so QbitAI infers roughly that figure. An industry source: "We pay one core scientist more than that a year."
- Market: over 100 robot retail kiosks or deployments in 30+ cities; at least ten-plus embodied-AI firms in retail; 17 companies' robots in real stores or sites per media reports.
- Beijing kiosk (about 10 m²): staffed 10:00-22:00 to restock, charge robots and report faults; a pick takes 20-plus seconds; misplaced or missing items stall the robot. A convenience-store robot works only at the counter with rigid-pack items; pick time fell from nearly a minute to ten-odd seconds after tuning. Dexterous hands were skipped on cost.
- Economics: a kiosk holds a few dozen SKUs against 2,000-3,000 in a convenience store. Reported revenue examples: 21,000 yuan in a May Day week for one shop; over 10,000 yuan a day at some sites at Spring Festival. One listed AI group's new businesses, including robot retail, were 6.0% of revenue.
- Interviewees call it proof-of-concept and marketing: robot makers need deployment sites more than retail needs robots; tourist sites tolerate faults and spread videos. No firm has shipped more than 100 robots into a single retail scenario. Ice cream takes 6 minutes.
- Requirements named for real retail: safety around children, site-specific customisation (uneven floors, non-standard shelves), robot cost below a human clerk, payback within a year. Industry estimate for replacing clerks: two to five years.

## How they got it

Company announcements and annual reports, visits to a Beijing kiosk and a convenience store, and interviews with a person close to Hongqi, a former UBTech sales manager, a retail-solution manager, a pharmacy-robot salesperson and a chief scientist. Revenue examples are from other media reports.

## Limits

- Most interviewees are identified by single names or roles only; companies are mostly unnamed.
- The project cost is an inference from one line in the annual report.
- Kiosk revenue figures are selected media examples, with no cost side.
- Library summary says 60-70% of buyers are tourism sites; the source says 60-70% of robot retail kiosks are placed at tourist sites and landmarks.

## Why it matters for companies

A clear account of the gap between a lab demo and a shop floor: mixed packaging, condensation, narrow aisles and customers break grasping systems that pass lab tests. For retailers and consumer-goods firms evaluating store robots, the useful tests are full staffing cost (people still restock, charge and watch), pick time, SKU range and payback period, not interaction quality. Hongqi's pivot to remote staffing is also the cheaper automation path for night trade today. The share-price jump on announcement is a reminder that robotics news can be priced in long before anything works.

## Key passages in translation

> 机器人的能力很有限，比如夏季玻璃瓶冷饮易结水珠，机器人拿不稳总摔碎

The robot's abilities were very limited: in summer, glass bottles of cold drinks gather condensation, and the robot could not hold them steadily and kept smashing them.

> 机器人更需要落地场景，而不是（零售）场景更需要机器人。

Robots need deployment sites more than (retail) sites need robots.

> 零售仓每天早10点至晚10点营业，需要有人一直值守。

The retail kiosk is open from 10 am to 10 pm every day and needs a person on duty the whole time.

Original: https://www.qbitai.com/2026/10/505955.html