Safety challenges for AI agents' ability to learn and act in desired ways in relation to biologically and economically relevant aspects. The benchmarks are implemented in a gridworld-based environment. The environments are relatively simple, just as much complexity is added as is necessary to illustrate the relevant safety and performance aspects.

Enables you to convert a PettingZoo environment to a Gym environment while supporting multiple agents (MARL). Gym's default setup doesn't easily support multi-agent environments, but this wrapper resolves that by running each agent in its own process and sharing the environment across those processes.

A reinforcement learning project implementing a Deep Q-Network agent that learns goal oriented navigation in a custom grid environment, with policy evaluation, visualization, and analytics.

Research-grade Reinforcement Learning framework for single-agent and multi-agent warehouse navigation using Deep Q-Networks (DQN), PyTorch, replay buffer, target networks, logging, and full test suite. Built for PhD-level RL and autonomous systems research.

Simulating the autonomous ship navigation in a gridworld.

Extended, multi-agent, and multi-objective (MaMoRL / MoMaRL) gridworld environments building framework based on DeepMind's AI Safety Gridworlds. This is a suite of reinforcement learning environments illustrating various safety properties of intelligent agents. It is made compatible with OpenAI's Gym/Gymnasium and Farama Foundation PettingZoo.

python package for fast shortest path computation on 2D polygon or grid maps

Experimental AlphaZero-style RL agent for optimizing strategies in the StarCraft II Arcade map 'New Random Tower Defense'.

A modular, extensible, entity-component-system (ECS) gridworld environment

An implementation of Value Iteration and Policy Iteration to solve a stochastic, grid-based Markov Decision Process (MDP), using the Gridworld environment.

Accelerated minigrid environments with JAX

𝙵𝚎𝚍𝚎𝚛𝚊𝚝𝚎𝚍 𝙼𝙰𝚁𝙻-𝙶𝚢𝚖 : We introduce a custom multi-agent reinforcement learning environment built with Gymnasium and Pygame, designed for evaluating federated RL (FRL) algorithms. The environment models a grid world where multiple agents—such as robots—navigate to accomplish spatially distributed tasks, like reaching delivery points.

RL Maze Project

Implementation of Q-learning to solve GridWorld

My first reinforcement learning model, designed to navigate through randomly positioned obstacles to reach a target.

Transforming a previous university project mini-game GridWorld into a proper Game using the Simple DirectMedia Layer and C++.

Simple Grid Environment for Gymnasium

R.L. methods and techniques.

Markov Decision Process: an agent has to navigate a grid in an optimal way (i.e., maximising rewards while minimising penalties)