Autonomous Mobile Robotics

The aim of our research is to reduce the need for large specially developed assembly systems by using small flexible robots. For this purpose, methods have to be developed to dynamically adapt existing mobile units to new tasks. In addition, methods must be found to control a group of mobile robots as one large handling system. The current focus is on the one hand on the modeling and control of the individual mobile manipulators and the development and selection of the required hardware. On the other hand, algorithms are being worked on that enable the control of the individual mobile robots as a robot network. This includes not only the control of the robot network but also its adaptation to the current environmental situation.  

In the field of cooperative component handling, the goal is to transport lightweight but oversized components using multiple serial robots. For this purpose, each robot arm is equipped with a gripper that establishes either a form or force-fit connection to the object. This makes it possible to lift and manipulate the object with multiple robotarms. However, this requires high-precision control of the robotic arms involved. If any robotic arm is incorrectly positioned or oriented, forces will be introduced into the object or the robotic arm joints. This would either lead to a deformation or destruction of the respective object.

In the video shown, such a control system can be seen with two robotic arms manipulating an elongated object. To enable this interaction of the robot arms, one robot arm is operated with an impedance control and the other robot arm with an admittance control. At the same time, a target trajectory for the object is specified, resulting in the shown motion.

The transport of large objects is not always possible using robot arms. Instead, a heavy component can be placed on several mobile platforms to be subsequently transported to a following production step. For both types of transport, it is necessary that the mobile platforms travel in a predefined formation that is not violated at any time. Again, any deviation in the position of a mobile robot can result in damage or destruction of the object. If a mobile platform deviates too much from the target position, the object being transported may fall to the ground, a robot arm may be overloaded, or it may be brought to its workspace limit.

Cooperative object transport requires many software modules from different disciplines to work seamlessly together. For example, mapping, localization, costmap, path planning and control are important. For the formation movement shown in the video, a costmap layer, global and local path planner were developed at the Institut of Assembly Technology. The motion is determined by specifying a target point in a recorded map, after which a motion for the formation is planned and executed by the mobile robots. The used formation consists of two real robots and one virtual robot, which hold a triangular formation during the movement.