The Robocup International annual event showcases the latest developments in robotics and artificial intelligence, featuring teams of humanoid robots competing in football.
Teams of scientists from all over the world participate in Robocup, working on the development of robots and algorithms. Before testing these systems in real-world conditions, most of them are first evaluated through simulations.
In a recent study, researchers from Dalhousie University and Memorial University of Newfoundland in Canada have presented pyrus basy, a new open platform for two-dimensional (2D) simulations of Robocup football. This platform, built on Python, provides a reliable means for specialists to train and test their algorithms. The study was published in the Journal of Robotics and Artificial Intelligence Research.
Dr. Aref Sayarech, one of the scientists involved in the project, commented, “We have created this basic code to simplify the work of researchers in this multi-user environment, allowing them to focus on their ideas rather than getting caught up in complex programming.”
Existing platforms for simulating soccer, such as Heliosbase and Cyrus2dbase, primarily utilize the C++ programming language. Despite its advantages, C++ can be more challenging to work with compared to Python and other languages. Recognizing this, Sayarech and his colleagues aimed to develop an alternative platform accessible to users with varying levels of programming expertise.
Sayarech explained, “We have designed Pyrus Base to be as straightforward as possible, enabling researchers to focus on their ideas without being constrained by complex mathematics and programming.”
Pyrus Base is a user-friendly and cost-effective platform for 2D simulations, allowing even beginners to test AI models for Robocup. The researchers hope that their platform will attract more scientists and engineers to participate in the Robocup community.
Highlighting the significance of AI in this field, Sayarech stated, “Artificial intelligence can play a significant role in this environment, particularly through the implementation of reinforcement learning methods.”
The base code for Pyrus is already available on GitHub and can be downloaded and used by robotics and scientific groups worldwide. In their future work, the researchers plan to further optimize the platform’s performance and provide reinforcement learning examples for beginners.
Dr. Sayarech added, “We are continually working to enhance Pyrus Base and expand its capabilities, aiming to make it even more effective in training a wider range of AI algorithms.”