Control of flexible mobile manipulators: positioning and vibration reduction using an eye-in-hand range camera

Abstract

In this paper, the positioning of a long flexible manipulator on a moving platform is investigated. The problem is to position the gripper at a requested relative distance in front of an object with unknown location. For this purpose, the gripper is equipped with a range camera giving the distance to surrounding objects within, ∼1% and with a sampling rate above 1 kHz. The range measurements are used in combination with internal angle measurements from joint encoders to estimate both the flexibility in the mechanical construction and the relative distance from gripper to object. This is solved satisfactorily by an extended Kalman filter (EKF). For the motion control of the manipulator, a time-scaled feedback controller is suggested. A fast inner loop is used to damp out oscillations and reject disturbances, both from the platform and the manipulator. An outer control loop, with a lower closed-loop bandwidth, then steers the gripper, based on the range measurements, to the requested final position in front of the object. This loop assumes a stationary and rigid platform and a rigid manipulator. At this moment, only simulations of a flexible manipulator on a rigid platform have been studied. However, the results show that the flexibility can be estimated from indirect measurements of the range to the object and the joint angles. Also, good damping and disturbance rejection are achieved, as long as the bandwidth of the actuators is sufficiently high compared to the oscillation. The use of range measurements of the surrounding objects makes the positioning task very robust against an uncertain platform position.