Enhancing RULER width collaborative diffusion for AI-driven games.
RULER Platform, Collaborative Diffusion, AI Pathfinding, Computational Thinking, Block-Based Programming, Educational Tools
The goal of this project is to integrate collaborative diffusion into the RULER platform to enhance AI pathfinding capabilities, making the tool more intuitive and engaging for young students learning computational thinking through game development.
The RULER platform lacks advanced AI pathfinding capabilities, limiting the complexity and engagement of the games that users can create, which becomes frustrating as they advance in knowledge and ambition.
We successfully integrated the Collaborative Diffusion AI code block into the RULER platform, enhancing game complexity and engaging students. User tests with 1st to 4th graders showed increased excitement but no significant improvement in understanding the shortest path concept. The new code block worked well without performance issues but highlighted areas for further educational improvement.
User Test:
During user tests, we assessed students' ability to complete a game incorporating Collaborative Diffusion, noting the ease or difficulty encountered. To evaluate the new code block's impact on excitement and engagement, we observed students, recording their frustrations, need for help, and moments of excitement. We conducted pre- and post-surveys to gauge understanding of Collaborative Diffusion. Students were shown a screenshot of a Pac-Man game and asked which direction the ghost should take to catch Pac-Man quickly, explaining their choice. The correct answer was to move left. The same question was asked after using the Collaborative Diffusion code block in their own game. In the workshop, students were introduced to the RULER platform and given step-by-step instructions to program their Pac-Man game. They added agents, implemented Pac-Man's movement, and initially made the ghost move randomly. Then, they replaced the random-movement block with a new AI-Pathfinding block to enable more intelligent movement.
User Test Results:
Students successfully integrated the Collaborative Diffusion code blocks into their Pac-Man games without any issues, and the algorithm worked consistently across all grid sizes and pursuer numbers. Students were enthusiastic about programming their games and were excited by the new intelligent ghost movement feature. However, the simplicity of changing only one block and the increased game difficulty with smarter ghosts led to some disappointment. Pre- and post-survey results showed no significant change in students' understanding of ghost movement direction. Those who initially answered correctly continued to do so, while others did not change their incorrect answers. Only a few students correctly explained the ghost's movement. An analogy comparing Pac-Man to cheese and the ghost to a mouse helped students understand the concept of Collaborative Diffusion. However, they did not pay much attention to the color gradient in the code block preview.
Prof. Dr. Alexander Repenning Chair of Computer Science Education Pädagogische Hochschule FHNW Windisch 5210 Switzerland Alexander Rapenning
Michelle Rüegg Basil Trueb
Martin Kropp martin.kropp@fhnw.ch
