Tracking of the crane load motion is important in order to achieve accurate control. This can be achieved by using marker based visual tracking. The work in this project has so far focused on improving the tracking of the crane load. In the future effort will probably also be put into improving crane control algorithms.
ArUco markers were used to facilitate tracking of the crane load. Read more about the use of ArUco markers here.
A method for calibration of the marker positions on the crane load and the mass position parameter used in the dynamic model was presented in the paper:
Myhre, Torstein A., and Olav Egeland. "Estimation of crane load parameters during tracking using expectation-maximization." American Control Conference (ACC), 2017. IEEE, 2017.
The crane load dynamical model in this paper is based on the Spherical Pendulum Dynamics model
The rotation of the crane load is represented by the Euler angles ϕ_x, ϕ_y and ϕ_z. The parameter θ_l represent the distance from the pivot point to the mass center of the crane load, g is the acceleration of gravity.
An accurate value of the parameter θ_l and the position of each marker on the crane load relative to the center of mass is required for accurate tracking. The method for estimation of these parameters described in the paper is based on Expectation-Maximization.
The method described below is presented in the paper:
Myhre, Torstein A., and Olav Egeland. "Collision detection for visual tracking of crane loads using a particle filter." Industrial Electronics Society, IECON 2016-42nd Annual Conference of the IEEE. IEEE, 2016.
The proposed method was developed for detecting collisions of the crane load during tracking. To test this method a laboratory experiment was performed using a UR5 robot manipulator as a crane simulator. A Prosilica GC 1020 ethernet camera was mounted on the end-effector in order to track the ArUco marker on the crane load.
The crane load was made part of an electrical circuit in order to verify the collisions the collisions detected by the particle filter algorithm. A probe was used to simultaneously close the electrical circuit and disturb the motion of the crane load. In this manner the collisions detected by the particle filter algorithm could be verified.
A video of this experiment is shown below.