Developing a control architecture for a vision based automatic pallet picking

Abstract

The goal of this project is to enhance the controlling performance of an articulated-frame-steering autonomous machine. The current system employs the vision sensor for the detection of the target object and controlling the machine movement by the data obtained by that. The problem we face is that at a far distance when location of the object is detected for the first time, the unreliable data specially the orientation of target will be acquired. So relying on this type of data for heading the machine can distract the whole system behavior, and also there should be a logic for switching between different states of machine in order to control the different stages of performance. In order to solve the problem, a smooth switching logic will be defined to control the machine. This switching logic should be in a case that operational coordinates with the visual servoing be synchronized and it should be planned to control robot degrees of freedom in each step. The MATLAB/Simulink is employed to execute the idea since this software is capable in the logic-based planning in a graphical way in stateflow. The results of the implemented method would be increasing the accuracy of picking the object and enabling the machine to have more space to have any needed maneuver, moreover the environmental disturbances will have the minimum effect on the final result.