Autonomous guidance of initial welding position with “single camera and double positions” method

Abstract

Purpose

Recognition and guidance of initial welding position (IWP) is one of the most important steps of automatic welding process, also a key technology of autonomous welding process. The purpose of this paper is to advance an improved Harris Algorithm and grey scale scanning method (GSCM) to raise the precision of image processing.

Design/methodology/approach

Through the configuration of “single camera and double positions,” a new set of image processing algorithms is adopted to extract feature points by using the pattern of rough location and subtle extraction, so as to restructure three‐dimensional information to guide robot move to IWP in the practical welding environment.

Findings

Experiments showed that mean square errors (MSEs) in X, Y, Z‐directions for both flat butt joint and flat flange are 0.4491, 0.8178, 1.4797, and 0.5398, 0.4861, 1.1071 mm, respectively.

Research limitations/implications

It has a limitation in providing guidance for only one step, and would be more accurate if fractional steps are adopted.

Practical implications

Guidance experiments of IWPs on oxidant tank's simulating parts are carried out, whose success rate is up to 95 percent and MSEs are 0.7407, 0.7971, and 1.3429 mm. It meets the demands of continuous and automatic welding process.

Originality/value

Improved Harris Algorithm and GSCM are advanced to raise the precision of image processing which influenced guidance precision most.

Citation

Wei, S., Ma, H., Lin, T. and Chen, S. (2010), "Autonomous guidance of initial welding position with “single camera and double positions” method", Sensor Review, Vol. 30 No. 1, pp. 62-68. https://doi.org/10.1108/02602281011010808