Biweekly Report for the Bot Project
Subject: Milestone I - Summary of Current Work, Further Implementation of Algorithms and Visualizations, Project Unification
Project Title: BotWheels
Report Number: 3
Date: November 11, 2024
Week Range: Weeks 5-6
1. Progress Overview
The overall engagement from the team has decreased in recent weeks. Currently, we are focusing on unifying the project approach and developing a simple CLI to improve usability. At the same time, we are continuing to refine the PPO and A2C/A3C algorithms, with particular attention to improving the reward module for better performance.
2. Team Member Contributions, Marks, and Linked Issues
| Team Member | Mark | Summary of Contributions | Linked Issues |
|---|---|---|---|
| Jakub Cisoń | 4 | Team management, leading conceptual work, preparing presentation for milestone 1. Limited availability led to slight delays in task delivery to other team members. | |
| Kacper Drozdowski | 3 | Delivered research on reward functions, though with significant delay from the declared date in report 2. Currently assigned to implement reward functions in the code. | - Research about different reward functions - Implement Additional Reward Functions |
| Mateusz Gościniecki | 4 | Continued work on the rocket-learn approach – backup approach. Moderate involvement with little produced code. If the approach does not meet expectations, it will be reassigned next week. | - Rocket Learn: Creating a Basic Training Loop vs Bot |
| Anna Ivanytska | 3.5 | Delivered event logger for visualization with delay. Currently working on more advanced approaches. Poor communication. | - Expansion of Event Logger - https://github.com/bot-wheels/bot-wheels-core/issues/50 |
| Igor Malkovskiy | 2 | Assigned task for training parameterization – no feedback over the last 2 weeks. | - Transition Training Parameterization from In-Code to Config File |
| Dawid Mielewczyk | 4.5 | Standardization of project work – code. Currently working on integrating CLI into the project. Technical support – review requests. | - Develop CLI for Training and Gameplay Modes -Improving Work Structuring - Concept for Standardizing Variable Naming, Code Documentation, ETC |
| Camille Nadir | 2 | Very poor communication over the last weeks. No solutions delivered. | - Diagnostic Matches for Bot Performance Evaluation - Prepare Bot Performance Metric Tool |
| Michał Pryba | 5 | Very high engagement in the project. Offering own ideas. Continued work on the A2C algorithm. Implemented reward module into RL algorithms. | - Implement A2C Algorithm with RLBot -Support for A2C/A3C Implementation |
| Wojciech Szamocki | 5 | Very high engagement in the project. Offering own ideas. Delivered data logging for metrics to files and identified the visualization topic in bot training. Currently working on heatmap for bot activity on the field. | - Heatmap of Bot and Opponent Activity -Save Visualization Data as CSV After Match Completion |
| Konrad Siemiątkowski | 4 | Delivered research on optimization of learning parameters for A2C. Currently working on improving data collection for A2C and integrating game events. | - Refine A2C Data Collection & Integrate Game Event Metrics - #41 research and optimization of learning parameters of a2c |
| Michał Zarzycki | 4.5 | Work on PPO. Delivered logs to Google Drive. Delivered sample bot game result - recording. | - Task: Fix Issues in CustomObsBuilder Class -Implement PPO Training Against Human Players |
3. Challenges and Problems Encountered
There have been communication issues with some team members, along with reduced overall engagement compared to the project's start. This has led to delays in delivering solutions. Additionally, some individuals have taken on significant responsibility for the team, resulting in a relatively large workload for them, but this has helped move the project forward.
4. Plans for the Next Period (Weeks 7-8)
In the next period, the goal is to deliver the CLI and a more effectively trained bot capable of consistently performing actions, such as scoring goals. The bot's performance will be evaluated based on predefined criteria leading to ensure it meets the expected outcomes. Initial measurements will be taken to compare the bot’s behavior against the assumptions made during the planning stages, providing a foundation for further refinement and adjustments.
5. Summary
Improved self-motivation within the team is expected to lead to better coordination and more consistent progress. With increased individual drive, the team members will be more proactive in completing tasks and communicating effectively, which should help streamline the development process. This motivation is crucial for ensuring that everyone stays focused on their responsibilities and contributes to the overall success of the project.