A dynamic model for GPS based attitude determination and testing using a serial robotic manipulator

Abstract

A computational algorithm is developed for estimating accurately the attitude of a robotic arm which moves along a predetermined path. This algorithm requires preliminary input data obtained in the static mode to yield phase observables for the precise, 3-axis attitude determination of a swinging manipulator in the dynamic mode. Measurements are recorded simultaneously by three GPS L1 receivers and then processed in several steps to accomplish this task. First, artkconv batch executable converts GPS receiver readings into RINEX format to generate GPS observables and ephemeris for multiple satellites. Then baseline vectors determination is carried out by baseline constrained Least-Squares Ambiguity Decorrelation (LAMBDA) method that uses double difference carrier phase estimates as input to calculate integer solution for each baseline. Finally, attitude determination is made by employing alternatively Least-squares attitude determination (LSAD) in the static mode and extended Kalman filter in the dynamic mode. The algorithm presented in this paper is applied to recorded data on Mitsubishi RV-M1 robotic arm in order to produce attitude estimates. These results are confirmed by another set of Euler angles independently evaluated from robotic arm postures obtained along the predefined trajectory.