Darmstadt Dribblers 2005: Humanoid Robot

This paper describes the hardware and software design of the two humanoid robot systems of the Darmstadt Dribblers. The robots are used as a vehicle for research in control of locomotion and behavior of humanoid robots with many degrees of freedom and many actuated joints. The Humanoid League of RoboCup provides an ideal testbed for such aspects of dynamics in motion and autonomous behavior as the problem of generating and maintaining statically or dynamically stable bipedal locomotion is predominant for all types of motions during a soccer game. A new modular software architecture has been developed for efficient and effective implementation and test of modules for sensing, planning, behavior, and actions. 1 RoboCup and aspects of robot motion and behavior The RoboCup scenario of soccer playing legged robots represents an extraordinary challenge for the design, control and stability of bipedal and quadrupedal robots. In a game, fast motions must be planned autonomously and implemented online which preserve the robot’s stability and can be adapted in real-time to the quickly changing environment. Existing design and control strategies for humanoid robots can only meet these challenges to some extent. During the nineties, both trajectory planning methods relying on nonlinear robot dynamics and model-based control methods have evolved into the state-of-the-art for developing and implementing fast and accurate motions for industrial manipulators. Successful control of the nonlinear robot dynamics is also the key to fast and stable motions of bipedal and quadrupedal robots. Many subproblems remain unsolved in fulfilling this objective. One aim of the Darmstadt Dribblers is to contribute to this ambitious goal by discussing fundamental principles and recent methods in the modeling, simulation, optimization and control of legged robot dynamics. Further aspects on these topics are described in [4]. For developing and testing the different modules involved in an autonomous behavior as required in the RoboCup scenario a tailored software architecture is essential. The underlying software and robot control architecture has to accomplish the demands appearing in a highly nonlinear physical dynamical system as a humanoid robot. A definition of the current RoboCup environment is described in the rules for Humanoid League [8]. 2 Technical Data of the Humanoid Robots Currently two different types of humanoid robots are used, which differ mainly in height: Mr. DD and Mr. DD junior, see Fig. 1. Mr. DD was first used in RoboCup 2004. It is based on a unique prototype of a biped walking machine which has been custommade for our purposes by iXs Research Corporation (http://www.ixs.co.jp). Due to several hardware and software modifications the walking motion could be improved and some autonomous behavior was developed.. Mr. DD junior is based on the robot kit KHR-1 by Kondo (http://kondo-robot.com). Both of them have legs with six degrees of freedom (DOF) and are equipped additional with a camera system to sense the environment. They have nearly the same kinematic structure. In Fig. 1 the arrangement of joints is presented for Mr. DD. The small robot lacks the waist joints and one of the neck joints. For more information about hardware modification see section 4. The technical data are given in Tab. 1. It is planned to add more platforms for humanoid robots in 2005 and 2006. Fig. 1. Humanoid robot systems Mr. DD (left) and Mr. DD junior (middle) in real life and kinematic structure of Mr. DD (right). 3 Global software concept and modular software structure To establish a communication between the different software modules of an autonomous robot, an object-oriented framework has been implemented. The main focus of the framework is to simplify the realization of a suitable robot-control-application for a programmer. The developer should not worry about the different operating systems present Robot system: Mr. DD Mr. DD junior Height: 68 cm 37.5 cm Width: 31 cm 18.5 cm Weight: 4.8 kg 1.5 kg Degrees of freedom: 24 in total with 21 in total with 6 in each leg, 4 in each arm 6 in each leg, 4 in each arm, 2 in waist, 2 in neck 1 in neck Sensors: 1 camera (Philips, ToUCam, 1 camera (HewlettPackard, resolution: 160 x 120) resolution: 160 x 120) 24 joint angle encoders, 3 force sensors in each foot Control frequency: 20 ms 40 ms Processor: NEC Vr4181A 133 MHz Intel PXA 255, 400 MHz Operating system: Linux Windows CE Network: Wireless LAN, LAN Wireless LAN, Bluetooth Power supply: on-board batteries on-board batteries Table 1. Technical data of both humanoid robots. in our robots (Linux, Windows) just as little as about threads and communication, but concentrate very exclusively upon the features and actions which should be carried out. Therefore one regards a highly simplified example of a standard robotic control as explained in Fig. 2. The rectangles mark here the functional components. The ellipses in contrast represent the exchanged information. Fig. 2. Example of simplified robotic control These functional components, which accomplish the data processing steps, are called modules in the further process. The goal of the framework is to handle the inter-module communication completely transparent for the modules. For developers it should be of no importance, whether another module, with which it exchanges data, runs in the same thread or on another computer. This gives the flexibility to relocate single modules, e.g. from the robot’s on-board computer to an external computer, without any relevance for the implementation of the modules. Furthermore this approach allows for an easy exchange of data between different robots acting in a team.