Visual Servoing via Nonlinear Predictive Control

Visual predictive control architecture based on image moments for manipulator robots

“…This type of local model based predictor was proposed by the authors of the paper in [8] for point features, and later the same local model based predictor was used for image features prediction [6]. The one-step ahead prediction of the image moments evolution, when the plant model is considered, can be computed using (17) and the discrete model (15) of the VCMD, resulting:…”

“…Considering the direct dynamic model of the robot, its joints and torques limits, the camera projection model and the visibility constrains, an IBVS structure based on nonlinear predictive control was proposed in [5]. In [6], image-based visual servoing is addressed via nonlinear model predictive control. All the above techniques are using point features in order to generate the predictive control law and, in the same time, a cost function is constructed by comparing the future behavior of the visual feature, behavior obtained from the predictor, directly with the desired configuration.…”

Multi-Rate Visual Feedback Robot Control

“…It is interesting to remark that errors which makes lower Z value affect more to the controller robustness than vice versa due to that the model is based on the image Jacobean, Allibert et al (2010). This could be different if other features are chosen, as it is well known by researchers in this field.…”

“…In Allibert et al (2010) visual servoing task is formulated into a nonlinear optimization problem in the image plane. The authors claim that with predictive control it is easy to take into account 2D and 3D constraints.…”

“…The well known local model based on the interaction matrix, Allibert and Courtial (2009); Allibert et al (2010), is used to perform the filtersmoother step. This model is linearized and "discretized" upon the sampling time 1 .…”