Multi-Objective Reinforcement Learning algorithms implementations.

Multi-objective Gymnasium environments for reinforcement learning

Extended, multi-agent and multi-objective (MaMoRL / MoMaRL) environments 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.

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.

[NeurIPS 2021] Accommodating Picky Customers: Regret Bound and Exploration Complexity for Multi-Objective Reinforcement Learning

Multi-Objective Multi-Agent RL with non-linear utility functions

A multi-objective reinforcement learning system based on Pareto Q-Learning to optimize traffic light control in SUMO, balancing travel time, emissions, and traffic flow.