The Cornell Robocup Team
This paper describes the Cornell Robocup Team, which won the RoboCup-2000 F180 championship in Melbourne, Australia. The success of the team was due to electro-mechanical innovations (omnidirectional drive and a dribbling mechanism) and the control strate
- PDF / 267,242 Bytes
- 11 Pages / 451 x 677.3 pts Page_size
- 30 Downloads / 168 Views
Sibley School of Mechanical and Aerospace Engineering Ithaca, NY 14853, USA [email protected] 2 Department of Theoretical and Applied Mechanics Ithaca, NY 14853, USA fnagy,[email protected]
3
Department of Computer Science Ithaca, NY 14853, USA [email protected]
This paper describes the Cornell Robocup Team, which won the RoboCup-2000 F180 championship in Melbourne, Australia. The success of the team was due to electro-mechanical innovations (omnidirectional drive and a dribbling mechanism) and the control strategies that rendered them e ective. As opposed to last year's \role-based" strategy, a \play-based" strategy was implemented, which allowed us to make full use of the robot capabilities for cooperative control. Abstract.
1
Introduction
The RoboCup competition is an excellent vehicle for research in the control of complex dynamical systems. From an educational perspective, it is also a great means for exposing students to the systems engineering approach for designing, building, managing, and maintaining complex systems. In an e ort to shift the current emphasis of the competition away from simple strategies to more complicated team-based strategies, the main emphasis of this year's team was to play a controlled game. In other words, in a game without ball control, e ective strategies essentially consist of overloading the defensive area during a defensive play (the so called \catenaccio" in human soccer, a strategy that is very e ective, if not extremely dull and frustrating for the spectators), and shooting the ball towards open space or the goal area in the opponent's half during o ensive plays. This was, in fact, the simple role based strategy adopted by our championship team in 1999, which was shown to be extremely e ective. In order to bring coordination and cooperation to the RoboCup competition, the Cornell team developed two electro-mechanical innovations (omni-directional drive and dribbling, described in Section 2) and the associated control strategies that rendered them e ective (described in Sections 3 and 4). The key system concept employed by the Cornell RoboCup team is that of hierarchical decomposition, the main idea of which is depicted in Figure 1. P. Stone, T. Balch, and G. Kraetzschmar (Eds.): RoboCup 2000, LNAI 2019, pp. 41-51, 2001. c Springer-Verlag Berlin Heidelberg 2001
42
Ra aello D’Andrea et al. DESIRED FINAL POSITIONS AND VELOCITIES, TIME TO TARGET
DESIRED VELOCITIES
TRAJECTORY GENERATION
STRATEGY
LOCAL CONTROL
FEASIBILITY OF REQUESTS
Fig. 1.
Hierarchical decomposition
At the lowest level is the local control, which physically occurs on the robots; local feedback loops regulate the wheel velocities about the desired wheel velocities commanded by a centralized computational unit (a workstation). The desired wheel velocities are generated by the Trajectory Generation block; these velocities are generated in such a way to ensure that the robots are physically capable of executing the maneuvers (as limited by power, friction, and stability constraints). The Trajec
Data Loading...
