# Real time trajectory planning in 3D for crane operations

Cranes have become widely used in recent decades. Cranes are used in factories, construction sites, shipyards, harbors and at various military installations to simplify the movement of large payloads. They are controlled by operators who have many years of experience, practice and they can move the load to the specified destination, avoiding the collision with other obstacles. This of course requires a planned trajectory and precise control of the crane.

In my thesis I worked on developing a crane control system that would be able to find a minimal path whilst avoiding other obstacles, and the mechanics of the crane is controlled so that the load can move in the trajectory. The crane system operates in a simulation environment. This required a three-dimensional mathematical model of the overhead crane and a friction model to get closer to the reality. I used the MATLAB/Simulink environment for the implementation and MuPAD for the calculations which is a symbolic tool in MATLAB. I have created a three-dimensional virtual model of the crane so that my work and the subsequent tests can be more spectacular. To create the virtual model, I used the V-Realm Builder 2.0 program which is an inbuilt tool of MATLAB. The virtual model has been implemented in the Simulink environment and the outputs of the dynamic system have been used for the proper animation.

After the implementation of the system, two trajectory planning algorithms were tested. The first one is a primitive algorithm that finds an optimal plane in the discrete workspace and in that plane finds the shortest path. The second one is a potential field trajectory planning algorithm. The calculations of the trajectory is not required to be executed in real time, but of course the calculation time must be the shortest possible. The planning algorithms have to provide a path in which a payload can reach its destination departing from its given start location. To follow the trajectory the control of the crane must be in real time. I solved tracing with the help of forces defined at certain times.