Position Error Modeling for Automated Construction Manipulators

Abstract

Hydraulically actuated construction equipment is rapidly being retrofitted with robotic control capabilities by several major manufacturers. However, position control errors caused by several factors are significant in these types of construction equipment. Errors are amplified if the manipulator and its operator must measure and locate objects in the equipment’s fixed reference frame. Both mechanistic and statistical approaches to correcting position errors are possible. A statistical approach is reported here that is validated based on experiments with a computer-controlled large-scale manipulator (LSM). The LSM is sufficiently representative of several types of construction equipment to be able to serve as a general test bed. In the regression analysis, three factors which are measurable in real time: distance, hydraulic pressure, and payload, were varied to determine their influence on position errors in the LSM. It was shown that with an integrated multivariable regression model, about 30% of the mean positioning error of the LSM can be reduced without the use of fixed external reference systems. The model can be implemented as simple, real-time regression equations.