Volume 32 Issue 1 - January 3, 2020 PDF
Intelligent Cooperative Object Transportation for Multiple Omnidirectional Automated Guided Vehicles
Yen-Chen Liu*, Ji-Xian Peng, Firhan Huzaefa
Department of Mechanical Engineering, College of Engineering
Firhan Huzaefa and Yen-Chen Liu, “Centralized Control Architecture for Cooperative Object Transportation using Multiple Omnidirectional AGVs,” IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Macau, Nov. 2019.
Jih-Sien Peng and Yen-Chen Liu, “Towards Cooperative Transportation of Multiple Mecanum-Wheeled Automated Guided Vehicles,” ASME Dynamic Systems and Control Conference (DSCC), Utah, USA, Oct. 2019.
Font Enlarge
Multi-robot systems have received considerable attention from researchers in the past decades. Inspired from the cooperation abilities in small living organisms like ants, birds, and fishes, researchers have been trying to bring out the same level of intelligence in multi-robot systems which is visible from the nature. The concept of multi-robot system have huge potential to be applied in industry. In practice, various loading weight and merchandise sizes lead to difficulties in storages and transportation within warehouses and industrial environment. In this research, we develop control architecture for cooperative transportation without physical link between automated guided vehicle (AGV). This will enable AGVs to contract and expand to different formations depending on object sizes and weight. Consequently, the variation of transporting different types of objects can be solved by adjusting the number and formation of AGV in the system.

The proposed system contains four parts, which are

1. Design and manufacturing of the omnidirectional AGV
The traditional AGVs using differential driving wheels is difficult to implement the concept of cooperative transportation. Thus, in this research we propose a novel design of AGVs by using mecanum omnidirectional wheels with a rotatory platform to provided flexibility of object transportation with various formations and trajectories.
Figure 1. Design of the proposed omnidirectional AGV

2. Synchronized control for multiple omnidirectional AGVs
If AGVs are controller to accomplish cooperative transportation individually, the uncertainties and tracking errors within one of the AGVs would lead to poor performance of the entire system and even mission failed. Therefore, we utilized the idea of synchronization control for networked dynamical system to couple the motion of all AGVs. Hence, the relative distance and orientation are synchronized so that the performance can be guaranteed.
Figure 2. Cooperative transportation of multiple AGVs with synchronized control

3. External force estimation for AGVs in cooperative transportation
In the design of control algorithms for cooperative transportation, the external force from the transported object is necessary to ensured system performance. However, the requirement of force sensors are expensive and inefficient in practice. Therefore, we present an estimation approach to utilize adaptive control, robust control, and neural network techniques to obtain the distributed force from the object to each of the AGVs. So that the force can be considered in the proposed controller to improve transportation performance.
Figure 3. Control flow of the proposed adaptive, robust, and neuran network estimation for external force in the proposd system

4. Trajectory generation of cooperation transportation
Although the proposed control algorithms can ensure both tracking performance and uncertain object information in cooperation transportation, an accessible and feasible trajectory and orientation are important in implementation. In this research, we utilize rapid-exploring random trees to obtain the desired trajectory of the transportation mission by given only the initial and final position and orientation. This method can generate a collision-free trajectory for the multiple AGVs system to transport the object successfully.
Figure 4. Desired trajectory and orientation for cooperative transportation generated from rapid-exploring random trees
< Previous
Next >
Copyright National Cheng Kung University