CooPick: Kollaborative Roboter-Roboter-Mensch Interaktion beim Fruchtauflegen

Problem Formulation

Depending on the flexibility and capacity requirements of a fruit-packaging station, the fruit-sorting and packing are so far carried out either manually or fully automated. Although the fully automated packaging systems are characterized by a high packing throughput, they are very cost-intensive and regarding the fruit varieties inflexible. The aim of this project is to develop a flexible and scalable robot system, which can be integrated into existing manual processes and is able to perform task support in collaboration with the workers. In this sense, several lightweight robots are to be used, which communicate with each other and coordinate their actions and movements. Furthermore, the robots are supported by a camera system and must be able to collaborate with the workers as part of a collaborative work organization. With robots and workers sharing a common workspace, special safety requirements and collision avoiding strategies for the robots have to be considered in order to achieve maximum efficiency, safety, and reliability.

 

Solution Approach

For a time-optimal robot task assignment and execution, we suggest a hierarchical control structure consisting of two-layer optimization-based control policies. In the upper layer, scheduling, a discrete optimization problem is used to compute the optimal allocation of the tasks to the robots (resources) by minimizing the total Euclidian distance covered by the robots' end effectors. The scheduler computes feasible final robot configurations for the underlying MPC-based trajectory planning layer. The MPC layer generates collision-free robot trajectories by minimizing the execution time of the robot's tasks and is therefore referred to as time-optimal Model Predictive Control. To this end, we have developed an efficient formulation of a collision-avoiding strategy based on a dynamic robot geometry approximation and tangent separating planes in conjunction with velocity constraints.  The generated optimal trajectories consisting of position, velocity, and acceleration data points are sent to the Robot controller in real-time using Robot Operating System (ROS) and a velocity control interface.

 

Project Goals

  • Mathematical modeling of the robot task allocation and scheduling as well as their coupling with the continuous robot dynamics.
  • Design and implementation of reliable consensus-based algorithms for cooperative selection of non-stationary objects.
  • Model based planning and control of cooperative collision-free trajectories for multiple robots sharing a common workspace.
  • Design and implementation of distributed optimization techniques considering the interaction between control and communication in order to fulfill the hard real-time requirements.
  • Collaborative robot-human interaction involving external high-resolution camera information.
  • Development of an experimental setup consisting of a conveyor belt, two lightweight robots and a camera system in order to test the developed algorithms.

 

CooPick Scene

Keywords

  • Cooperative Robotics
  • Model Predictive Control
  • Scheduling
  • Optimization and Control

 

Funding

Projektträger: AiF-Projekt GmbH

 

Time span

Jan 2018 – May 2021

 

Project Partners

Research: BIBA GmbH

Industry: MWZ Steuerungstechnik

 

Contact

Prof. Dr.-Ing. Naim Bajcinca
Gottlieb-Daimler-Str. 42
67663, Kaiserslautern
+49 (0)631/205-3230
naim.bajcinca(at)mv.uni-kl.de