ORIGINAL RESEARCH article
Front. Mech. Eng.
Sec. Mechatronics
Multi-State Robot Posture Detection Method Based on Joint Optimization of Target-Key Point Detection
Provisionally accepted
- Shenzhen Institute of Information Technology, Shenzhen, China
Select one of your emails
You have multiple emails registered with Frontiers:
Notify me on publication
Please enter your email address:
If you already have an account, please login
You don't have a Frontiers account ? You can register here
Abstract: Industrial robots are core equipment in intelligent manufacturing, and their posture detection accuracy directly determines their high -precision operation capabilities. Addressing the problems of existing visual inspection methods, such as focusing on a single robot state, difficulty in balancing real-time performance and accuracy, and uneven optimization of multi-joint angle errors, this paper proposes a multi-state robot posture detection method based on joint optimization of target and keypoints. This method uses an improved cross-stage local network as the backbone network, simultaneously outputting target detection and keypoint detection features, and achieving dual-task collaborative optimization through weighted joint loss. Simultaneously, a multi-state adaptive module is designed to determine the state through motion vectors from adjacent frames, compensate for motion ambiguity using directional fast rotation binary robust independent basic feature matching, and correct viewpoint deviations using posture correction parameters. Results reveal that the introduced model obtains an average target detection accuracy of 94.73% at an intersection-union threshold of 0.5:0.95, a correct keypoint detection probability of 98.7% at a keypoint detection threshold of 0.2, and average absolute errors of joint angles of 7.2°, 8.1°, and 9.5° in fixed, moving, and unknown states, respectively, with an inference speed of only 43 frames per second. The method can improve detection performance in multi-state scenarios, providing technical support for complex operations of industrial robots.
Keywords: Joint optimization, Key point detection, Multi-state robot, object detection, Robot pose detection
Received: 10 Dec 2025; Accepted: 02 Feb 2026.
Copyright: © 2026 Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
* Correspondence: Xiaoli Zhang
Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.