The Carologistics RoboCup Logistics Team 2013

In this team description paper, we outline the approach of the Carologistics team with an emphasis on the high-level reasoning system. We outline the hardware modifications and describe our software systems and describe our efforts towards a fully referee box. The team members of the 2013 team are Andre Burghof, Daniel Ewert, Alexander Ferrein, Bahram Maleki-Fard, Victor Matare, Tobias Neumann, Tim Niemueller, Florian Nolden, Sebastian Reuter, Johannes Rothe, Richard Schulz, Alexander von Wirth, Frederik Zwilling. 1 The Carologistics Robotino Robots The basic platform employs omni-directional locomotion, features twelve infrared distance sensors and bumpers mounted around the base, a CruizCore gyroscope, and a webcam facing forward. The Carologistics Robotinos have an additional omni-directional camera system as shown in Figure 1, taken from the AllemaniACs’ former middle-size league soccer robots [1], which allows for a 360◦ view around the robot. It is used to detect pucks around the robot. The webcam is used for recognizing the signal lights of the production machines. An additional Hokuyo URG laser scanner is used for collision avoidance and self-localization. In 2013, the robots will be outfitted with an additional laptop on the robot. 2 Middleware Concepts: Deploying Fawkes and ROS The software system of the Carologistics robots combines two different middlewares, Fawkes [2] and ROS [3]. This allows us to use software components from both systems. The overall system, however, is integrated using Fawkes. Adapter plugins connect the systems, for example to use ROS’ navigation and 3D visualization capabilities. Most of the functional components are implemented in Fawkes. For example self-localization is done using Adaptive Monte Carlo Localization. From ROS we use the locomotion package (move base) which implements a dynamic window approach for local motion and collision avoidance and a Dijkstra search for a global path. The behavior components have been developed on top of Fawkes, but could easily be used in ROS. For computational and (a) Modified Robotino of the Carologistics RoboCup team. (b) Image from the directed camera detecting the light signal of a machine. (c) Image from the omnidirectional camera. (d) Visualization of the scene in Rviz Fig. 1. Carologistics Robotino, sensor processing, and visualization energy efficiency, the behavior components need to coordinate activation and deactivation of the lower level components to solve computing resource conflicts. The behavior components are described in more detail in Section 3.2. Next, we briefly describe the task coordination components. 3 High-level Decision Making and Task Coordination Task coordination is performed using an incremental reasoning approach [4]. In the following we introduce the rule-based production system CLIPS, describe the behavior components, and briefly describe the reasoning process in two particular situations from the rules in 2012. 3.1 CLIPS Rules Engine CLIPS is a rule-based production system using forward chaining inference based on the Rete algorithm [5]. The CLIPS rule engine [6] has been developed and CLIPS Agent Sim World