Changan Automobile Launches First Visual AI Glue-Brushing Robot at Liangjiang Plant
中国长安汽车首例涂装视觉 AI 刷胶机器人上线，风险识别检出率与报送准确率均超 99.5%
Changan Automobile announced the recent launch of its first visual AI glue-brushing robot for painting operations at the Liangjiang factory. The system utilizes a deep convolutional neural network (CNN) to identify optimal gluing points and standardize trajectories, addressing challenges posed by complex body shapes and model variations. After rigorous validation, the robot achieved a risk detection rate and reporting accuracy both exceeding 99.5%, shifting quality control from post-process inspection to real-time process interception. This development eliminated bottlenecks in the PVC coating area, increasing line throughput by approximately 8%.
Source: IT Home — Robotics · Read original article ↗
Article text · Machine translation into English
IT Home November 5 news, Changan Automobile announced today that recently, China Changan Automobile's first painting visual AI glue-brushing robot was officially launched at Changan Automobile's Liangjiang Factory.
IT Home learned that the spatial structure of the undercoating station in the painting workshop is compact, with large differences in models and complex process brushing points, making the calibration of robot glue-brushing head points and the correction of operation trajectories extremely difficult.
Affected by factors such as differences in body and chassis shapes and model adaptability, it is very difficult to match the optimal process standards for the brushing head's operation angle, fitting distance, and walking trajectory in one go. In the painting undercoating PVC process, how to make the robot complete glue brushing accurately and uniformly like an experienced master has always been a recognized technical difficulty in the industry.
The official statement said that the Changan Automobile PVC bottleneck station efficiency improvement breakthrough team carried out a special breakthrough. The team introduced an AI vision platform and built a large model detection system based on Deep Convolutional Neural Networks (CNN), and finally, the team accurately locked in the optimal glue-brushing points and standardized operation trajectories, effectively adapting to the undercoating glue-brushing process requirements of all models.
After repeated verification, the coating visual AI glue-brushing robot has delivered a report card. The bottleneck in the PVC area was completely cleared, and the production line JPH increased by approximately 8%, significantly releasing capacity; AI vision achieved real-time online detection of the glue-brushing status, with single inference response reaching the millisecond level, The risk identification detection rate and reporting accuracy both exceeded 99.5%, achieving a shift from “post-event oversight” to “process interception.”
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
Reported accuracy
99.5%
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
What remains unknown
Not established in the collected evidence: Environment, Control, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
IT之家 11 月 5 日消息，长安汽车今日宣布，近日，中国长安汽车首例涂装视觉 AI 刷胶机器人在长安汽车两江工厂正式上线。
Open source S2
团队引入 AI 视觉平台，基于深度卷积神经网络（CNN）搭建大模型检测系统，最终团队精准锁定最优刷胶点位与标准化作业轨迹，有效适配全线车型的底涂刷胶工艺需求。
Open source S7
中国长安汽车首例涂装视觉 AI 刷胶机器人上线，风险识别检出率与报送准确率均超 99.5%
Open source S1
PVC 区域瓶颈彻底打通，产线 JPH 增幅约 8%，产能得到大幅释放；AI 视觉实现实时在线检测刷胶状态，单次推理响应达到毫秒级， 风险识别检出率与报送准确率均超过 99.5% ，实现从“事后把关”转向“过程拦截”。
Open source S9
Source：IT Home — Robotics · ithome.com