Joint-Space Kinematic Model for Gravity-Referenced Joint Angle Estimation of Heavy-Duty Manipulators

A gravity-referenced joint angle estimation, which acts in joint space, is proposed for multiple-degree-of-freedom hydraulic manipulators. As a novelty, the estimation pairs up inertial units across a section of an open kinematic chain. In this way, the three-axis linear accelerometers and the three-axis rate gyros provide a drift-free solution for observing the motion state of a rotary joint connecting two links without relying on the full forward kinematics. For a low-noise, low-delay estimate, the linear accelerations acting on the paired inertial units are modeled and robustly combined with the principles of complementary and Kalman filtering. In pick-and-place experiments with a serial-link manipulator on a multiton, off-road forestry vehicle, joint angle sensing error of less than +/- 1 degrees was achieved in spite of the dynamic interaction between the vehicle base and the terrain. Furthermore, the kinematic modeling's ability to compensate for the nonplanar, coupled 3-D linkage motion is studied for one-and two-axis rate measurements along with Cartesian path tracking. This gives new insights from the typical planar kinematic models and heavy-duty control viewpoints.