Lunar lander dqn github This is an implementation of DQN, DDQN, DDPG and TD3 on Lunar Lander environment from OpenAI Gym. The LunarLander-v2 environment in OpenAI Gym was used as the testing environment. Q-learning agent is tasked to learn the task of landing a spacecraft on the lunar surface. py at master · wtcherr/lunar-lander-dqn This repository contains an implementation of Deep Q-Learning (DQN) to solve the Lunar Lander environment using PyTorch and OpenAI Gym. Contribute to SNCA-24/LunarLander_v3_Benchmarking_DQN_vs_PPO development by creating an account on GitHub. This is the coding exercise from udacity Deep Reinforcement Learning Nanodegree. Contribute to srsapireddy/lunar_lander_using_DQN development by creating an account on GitHub. [1]. The Lunar Lander environment involves guiding a spacecraft to land on the surface of the moon safely. Leveraging the power of PyTorch, we've trained a sophisticated Deep Q-Learning model capable of executing precise landings between flag poles in a challenging virtual lunar landscape. DQN Solution for LunarLander-v3 (Gymnasium) | Reinforcement learning agent using Deep Q-Network (DQN) to land a lunar module safely. py at master · wtcherr/lunar-lander-dqn About Trained a Deep Q-Network (DQN) agent to land safely in the Lunar Lander environment using reinforcement learning. In this repository we implement an agent that is trained to play the game lunar lander using i) an actor-critic algorithm, and ii) a (double) deep Q-learning algorithm. More information is available on the OpenAI LunarLander-v2, or in the Github. This project utilizes a Deep Q-Network (DQN) to train the agent, leveraging PyTorch for the neural network implementation and Gymnasium (a fork of OpenAI Gym) for the simulation environment. Contribute to bmaxdk/OpenAI-Gym-LunarLander-v2-Double-DQN development by creating an account on GitHub. In this notebook, we will explore the implementation of a Deep Q-Learning (DQN) agent to navigate Gym's Lunar Lander environment. Each leg's ground contact nets +10 points. Contribute to djswanevelder/DQN_lunar_lander development by creating an account on GitHub. This project serves as a hands-on reinforcement learning experiment for solving continuous control tasks using deep Q-learning. Implementation and comparison of four Deep Q-Learning variants—Classic DQN, Double DQN, Dueling DQN, and Dueling Double DQN (D3QN)—for the Lunar Lander-v3 environment using PyTorch. Double DQN is a way of improving the DQN method which seeks to solve the over estimation problem occured in the classical DQN. Inspired by SpaceX's rocket landings, this AI agent learns to control a lander and successfully land it in a simulated environment. - dqna64/lunar-lander-dqn May 7, 2021 · In this post, We will take a hands-on-lab of Simple Deep Q-Network (DQN) on openAI LunarLander-v2 environment. (Alternatively, if you'd prefer to explore a complete implementation, enter the solution/ folder, and run the code in Deep_Q_Network_Solution. This is a simple DQN implementation to OpenAI/Gym/Box2d LunarLander-v2 using the DI-engine library and the DI-zoo. If the agent just lets the lander fall freely, it is dangerous and thus get a very negative reward from the environment. LunarLander_DQN Reinforcement Learning code applied to the Lunar Lander problem using MATLAB. Stable Baselines Integration This repository utilizes Stable Baselines, a set of reliable implementations of reinforcement learning algorithms. The Lunar Lander is a classic rocket OpenAI Gym's LunarLander-v2 Implementation. An additional reward of -100 or +100 points is received if the lander crashes or comes to rest respectively which also leads to end of the episode. Lunar Lander v3 implementation using DQN. Apr 20, 2025 · Lunar Lander v3 implementation using DQN. Instability The Lunar Lander environment is a well-known problem in reinforcement learning where an agent must learn to control a lander to safely touch down on the surface of the moon. Approach to resolve the task To resolve the task has been used a deep Q-network (DQN), implemented by Keras framework. Lunar-Lander-DRL-Dueling-DoubleDQN (D3QN) This is a Deep Reinforcement Learning solution for the Lunar Lander problem in OpenAI Gym using dueling network architecture and the double DQN algorithm. The agent is rewarded for smooth landings and penalized for crashing or flying out of bounds. LunarLander-DQN is a Python project that implements a Deep Q-Network (DQN) algorithm with extension for training an agent to land a lunar lander spacecraft. May 12, 2019 · An episode always begins with the lander module descending from the top of the screen. DQN was chosen as the main algorithm, seeing that it had succeeded in solving problems with similar environment setup. The project aims to achieve high The code describes my implementation of a reinforcement learning Agent (here, a deep Q-learning agent) to solve the open ai environment “LunarLander-v2” with discrete values. The agent is trained using a reinforcement learning approach where it learns to map states to actions in order to maximize cumulative rewards. Here is a video of a trained agent playing the game: video. Implementation of DQN in OpenAI Gym LunarLander-v2 discrete environment. ca) This file contains information on my implementation of DQN in the LunarLander-v2 environment. Experience replay to stabilize training. About 利用DQN将着陆器成功降落在月球表面 (Lunar Lander)/Using DQN to land the lander on the surface of the moon (Tensorflow-keras) Contribute to MorningStarTM/DQN-Lunar-Lander development by creating an account on GitHub. The project includes visual demonstrations of the lunar lander's trajectory, control adjustments, and final touchdown, emphasizing the effectiveness of RL in solving real-world-inspired The DQN agent is trained to solve the Lunar Lander environment provided by OpenAI Gymnasium. The goal is to successfully land a spacecraft on the ground. Current Projects Lunar Lander RL Practices: Explore different RL techniques applied to the Lunar Lander problem, from DQN to D3QN and beyond. Features: Custom neural network architecture for Q-learning. The agent is trained to optimize its landing strategy through reinforcement learning, specifically using the Deep Q-Network (DQN) approach. CartPole RL Practices: A collection of experiments focused on the classic Cart Pole problem using various RL algorithms. If the agent does not land quickly enough (after 20 seconds), it fails its 🚀 Lunar Lander DQN - Deep Reinforcement Learning. For each time step, there are four discrete actions Deep Q-Learning agent trained to land a Lunar Lander using Gymnasium 🚀 Includes training, evaluation, plots, and video playback 🎥 - sgrtrigg1/lunar-landar-dqn 🚀 Lunar Lander DQN - Deep Reinforcement Learning. Implementation References OpenAI baselines Reinforcement Learning (DQN) Tutorial Solving The Lunar Lander Problem under Uncertainty using Reinforcement Learning A simple PyTorch implementation of the Deep Q-Learning algorithm to solve Lunar Lander environment from Gymnasium. This project implements a Lunar Lander simulation using Deep Q-Learning (DQN). The agent has to learn how to land a Lunar Lander to the moon surface safely, quickly and accurately. The environment used is the OpenAI gym LunarLander-v2 environment. Deep Q-learning for lunar lander. VI. Contribute to SaintJeane/lunar-lander-dqn development by creating an account on GitHub. mp4 We use the lunar lander implementation from gymnasium. Implements experience replay, target networks, and training visu This is an implementation of DQN, DDQN, DDPG and TD3 on Lunar Lander environment from OpenAI Gym. RL agents trained to solve Open AI's Lunar Lander Gym - DQN - johngroves/lunar_lander DQN Solution for LunarLander-v3 (Gymnasium) | Reinforcement learning agent using Deep Q-Network (DQN) to land a lunar module safely. This report briefly introduces the Lunar Lander Environment from OpenAI Gym. It also includes an agent learns to play the Lunar-Lander game using Prioritized Experience Replay (PER). Contribute to ROCchender/LunarLander development by creating an account on GitHub. The best model and optimal hyper parameters found by a grid search, running lunarlander. 🤖🚀 - snowieeeee/Lunar-Lan The DQN agent is trained to solve the Lunar Lander environment provided by OpenAI Gymnasium. This project investigates the effectiveness of different Deep Reinforcement Learning (DRL) algorithms in solving the Lunar Lander control problem. Lunar_lander_DQN Training an agent with DQN to efficiently land the lunar rover using openai gynasium box2d env. The objective is to safely land a spacecraft between two flags using a variety of controls. Contribute to dcbothra/lunarlander_dqn development by creating an account on GitHub. This task is considered solved when Neural network controlling a simulated spacecraft learning to safe land on a surface using reinforcement learning techniques. ipynb. Normally, LunarLander-v2 defines “solving” as getting an average reward of 200 over an average of 100 consecutive episodes. Sam Weinberg (sam. In Double DQN, the target value for the main network An implementation of Double Deep Q-Network (DDQN) to solve the game Lunar Lander. Contribute to ZiruiXu/lunar-lander-dqn development by creating an account on GitHub. DQN is a popular reinforcement learning algorithm that combines Q-learning with deep neural networks. Fixed Q-target : separate local & target networks Experience Replay : Having a buffer of (state, action, reward, next_state, done) tuples to sample from Double DQN : using target network to evaluate the model- when choosing action maximizing action-value function ε-greedy Policy : choosing non-greedy action with probability = ε (starts at 1 and decays to 0 each episode) The episode finishes if the lander crashes or comes to rest, receiving an additional -100 or +100 points respectively. Reinforcement This repository contains a Deep Reinforcement Learning agent utilizing PyTorch to solve the LunarLander-v3 environment provided by OpenAI Gymnasium. learning reinforcement-learning openai-gym q-learning deep-q-network reinforcement ddqn double-dqn lunar-lander ddqn-lunar-lander Updated on Jul 25, 2024 Jupyter Notebook LunaLander is a beginner-friendly Python project that demonstrates reinforcement learning using OpenAI Gym and PyTorch. The environment for testing the algorithm is freely available on the Gymnasium web site (it's an actively maintained fork of the original OpenAI Gym developed by Oleg Klimov. The agent is based on the DQN algorithm with experience replay, target networks, and epsilon-greedy exploration strategy. In this article, we will cover a brief introduction to Reinforcement Learning and will learn about how to train a Deep Q-Network(DQN) agent to solve the “Lunar Lander” Environment in OpenAI gym. At each step, the agent is provided with the current state of the space vehicle which is an 8-dimensional vector of real values, indicating the horizontal and vertical positions, orientation, linear and angular velocities, state of each landing leg (left and A Deep Q-learning neural network is used to master Lunar Lander (OpenAi Gym) see another DQN I made with the Cartpole-v1 OpenAi Gym environment Aug 3, 2025 · LunarLander-DQN- Implementation and comparison of four Deep Q-Learning variants—Classic DQN, Double DQN, Dueling DQN, and Dueling Double DQN (D3QN)—for the Lunar Lander-v3 environment using PyTorch. - faraz1272/dqn-lunar-lander Contribute to VishalVShetty/Lunar_Lander_DQN development by creating an account on GitHub. In the Lunar Lander environment, DQN uses a deep network to approximate the Q-value function, which tells the agent how good or bad it is to take certain actions in specific states. In this project, I demonstrate the implementation and evaluation of Deep Q-Learning (DQN) and Double Deep Q-Learning (Double DQN) algorithms to solve the Lunar Lander environment from Gymnasium. But to improve the efficiency of the The Lunar Lander v2 environment is learnt using Deep Q Learning in Pytorch - mmahdim/LunarLander-DQN-DDQN Solving Lunar Lander with Deep Q-Network. The landing area is static and it is located always at the (0, 0) coordinates. This repository contains an implementation of the Deep Q-Network (DQN) algorithm and its variant DDQN to train agents to play the LunarLander-v2 environment in OpenAI Gym. DQN, SAC, DDPG, TD3, PPO and VPG implemented in pytorch. Deep Q-Learning (DQN) is a type of reinforcement learning (RL) algorithm. Soft update of target Q-network. However, DQN suffers from: Overestimation Bias: It tends to overestimate the action values. The goal of the task is to safely land a lunar lander ship within the landing area, the area between two yellow flags. This project solves a simplified version of a lunar lander problem under OpenAI Gym environment using Deep Q-Network (DQN). The agent learns to land the lunar module safely using reinforcement learning. ) After you are able to get the code working, try This project implements a Deep Q-Learning agent to successfully land a lunar module using the OpenAI Gym environment LunarLander-v3. CONCLUSION In this project, we successfully created a working agent that was able to navigate the Lunar Lander environment efficiently and provided a thorough comparison of model hyper-parameters. Contribute to rchen19/DQN_Lunar_Lander development by creating an account on GitHub. To begin, navigate to the exercise/ folder, and follow the instructions in Deep_Q_Network. About using DQN to control the lunar lander in gym environment The DQN also made use of an extremely large replay buffer (over 500,000 state-action transitions) and a target network that used hard-tau updates every 600th step along with annealed exploration. More info. Contribute to ranjitation/DQN-for-LunarLander development by creating an account on GitHub. The observation space of the environment consists of 8-dimensional states providing the following information about the spacecraft: x and y coordinates, x and y axes velocity The Lunar Lander environment simulates landing a small rocket on the moon surface. It utilizes the Gymnasium library for the environment simulation and Stable Baselines3 for reinforcement learning. May 28, 2025 · Contribute to DHANA5982/Gymnasium_Lunar_Lander_DQN development by creating an account on GitHub. This project uses Deep Reinforcement Learning to solve the Lunar Lander environment of the OpenAI-Gym - pramodc08/LunarLanderV2-DQN In this exercise, you will implement Deep Q-Learning to solve OpenAI Gym's LunarLander environment. DQN and variants SAC and PPO for the Lunar Lander Continuous environment by OpenAI - Rem4rkable/Solving-the-LunarLanderContinuous-v2 This is a solution to the Lunar Lander video game using Deep Q-Learning and OpenAI Gym - shperry03/DQN-Lunar-Lander 强化学习，在LunarLander任务环境上实现PPO算法、DQN算法以及DQN变体算法. About Reinforment learning solution using DQN algorithm for the Lunar Lander V3 environment from Gymnasium. Apply Double Deep Q Learning. Solving Lunar Lander with Deep Q-Network. May 1, 2025 · A Deep Q-Learning (DQN) agent trained to solve the LunarLander-v3 environment using PyTorch and OpenAI Gym. An implementation of Rainbow DQN on OpenAi's Lunar Lander Environment - Rainbow-DQN/README. The goal of this project is to train an agent using the DQN algorithm to land a spacecraft safely on a designated landing pad in the "LunarLander-v3" environment. pdf) for further readings and check out the training demo video at the end! May 3, 2020 · Lunar Lander Environment The state of a Lunar Lander environment has eight continuous values that represent the lander’s x and y position, it’s velocity, angular speed, orientation, and other. Implement Lunar-lander problem using DQN Contribute to Pawan4356/LUNAR-LANDER development by creating an account on GitHub. Contribute to drawar/lunar-lander-dqn development by creating an account on GitHub. The episode finishes if the lander crashes or comes to rest. in Data Science at University of Bath. Lunar Lander - Deep Reinforcement Learning, Noise Robustness, and Quantization Final project for EE460J Data Science Lab By Niruti Dhoble, Jianchen Gu, Indhu Gunda, Shreyas Kudari, Isaac Lee, Sophia Xu, and Kory Yang Built with: Introduction Deep Reinforcement learning is an exciting branch of AI that closely mimics the way human intelligence explores and learns in an environment. 🤖🚀 - snowieeeee/Lunar-Lan Implementation of a DQN for the Lunar Lander env. It was Contribute to DHANA5982/Lunar_Lander_DQN development by creating an account on GitHub. The input state of the Lunar Lander consists of following components: Horizontal Position Vertical Position Horizontal Velocity Vertical Velocity Angle Angular Velocity Left Leg Contact Right Leg Feb 8, 2025 · The Lunar Lander program in reinforcement learning with Deep Q-Networks (DQN) trains an agent to safely land a spacecraft on the moon by using a neural network to approximate Q-values and optimize 🚀 Lunar Lander DQN - Deep Reinforcement Learning. - lunar-lander-dqn/main. The Lunar Lander environment is based on the open source physics engine Box2D. The lunar lander can perform 4 actions. The goal is to land a simulated lunar lander safely on a designated landing pad. This project implements a DQN agent that learns to land a lunar module safely on a landing pad. This repository contains the code for the TU Delft project course "Bio-Inspired Intelligence and Learning for Aerospace Applications - AE4350" under the folder "Code". The solution was developed in a Jupyter notebook on the Kaggle platform, utilizing the GPU P100 accelerator. The Lunar Lander environment is a popular exercise for testing reinforcement learning algorithms, where Contribute to Bikxs/Lunar_Lander_DQN development by creating an account on GitHub. We sequentially implemented DQN, Double DQN, A2C, and SAC to observe how each algorithm improves upon its predecessors in terms of stability, learning speed, and final Contribute to SNCA-24/LunarLander_v3_Benchmarking_DQN_vs_PPO development by creating an account on GitHub. Hyperparameter tuning for Using Reinforcement Learning (DQN) to train a Lunar Lander for automated landing - npatel221/Lunar_Lander_RL Episodic DQN Algorithm: The notebook DQN Implementation implements the following algorithm for the descrete version the lunar lander problem: Jul 1, 2024 · DQN is a powerful reinforcement learning algorithm combining Q-learning and deep neural networks. Contribute to cranberrii/lunarlander-dqn development by creating an account on GitHub. DQN Implementation on Gymnasium's Lunar Lander. Added The goal of this project is to earn more than +200 reward on average over 100 trials in the game Lunar Lander. 🚀🕹️ This project applies Deep Q-Network (DQN) reinforcement learning to solve the LunarLander-v3 environment from OpenAI Gymnasium. While for the implementation Deep Q net for lunar lander. Tested Env: LunarLander-v2 and Pendulum-v0 Lunar Lander Deep Q-Learning This repository contains a Deep Q-Learning implementation for the LunarLander-v3 environment using PyTorch and Gymnasium. In the Lunar Lander environment, the agent's task is to DQN for LunarLander v2 Implementation of reinforcement learning algorithms for the OpenAI Gym environment LunarLander-v2 The episode finishes if the lander crashes or comes to rest. Normally, LunarLander-v2 defines "solving" as getting an average reward of 200 over an average of 100 consecutive episodes. But to improve the efficiency of the agent certain punishments were added in the reward section. Lunar Lander v3 (Gymnasium) Project This repository showcases an advanced deep learning project focused on developing an autonomous lunar lander using the Gymnasium environment. This is the GitHub home of Race Into Space, the computer version of the Liftoff! board game by Fritz Bronner. py script. {% cite mnih2013playing %} Here, we'll implement a simplified version of the DQN agent applied to the Gym Lunar Lander environment. This project implements a Deep Q-Network (DQN) to solve the LunarLander-v3 environment from OpenAI Gym. Environment is provided by the openAI gym 1 Base environment and agent is written in RL-Glue standard 2, providing the library and abstract classes to inherit from for reinforcement learning experiments. Implements experience replay, target networks, and training visualization. Implemented experience replay and a target network to stabilize training and improve performance. - abhaybodhe/Lunar-Lander-Environment. With DQN, it doesn’t really matter what they all are, I just need to know how many there are. - lunar-lander-dqn/dqn. We will use Google’s Deepmind and Reinforcement Learning Implementation for this. Aug 21, 2025 · This project implements a DQN agent that learns to successfully land a lunar module on the moon’s surface. md at master · kiranikram/Rainbow-DQN Lunar Lander v3 implementation using DQN. Contribute to akshitgaur2005/Lunar-Lander-DQN development by creating an account on GitHub. Lunar-Lander-DQN Algoritmo de Deep Q-Learn desenvolvido para a resolução de problema de aterrissagem de nave no jogo Lunar Lander Lunar Lander v3 implementation using DQN. Both DQN and PPO agents are used to train agents in the Lunar Lander environment. Contribute to twnkl2713/Lunar-Lander_DQN development by creating an account on GitHub. The agent is trained to control a lunar lander to safely land on a target platform. This project demonstrates advanced machine learning concepts including neural networks, experience replay, and epsilon-greedy exploration. 2016) is introduced and implemented with a detailed explanation. Key results demonstrate that both algorithms can solve the environment, with a detailed comparison between DQN and GitHub - Aalaa-A/lunar-lander-dqn: A simulation-based project showcasing the successful landing of a lunar module using reinforcement learning techniques. So, we turn to Function Approximations - An idea central to Contribute to Malhar124/lunar-lander-dqn development by creating an account on GitHub. 🚀 Lunar Lander DQN - Deep Reinforcement Learning. Also, a study has been done to learn how the hyperparameters of the Neural Network afect the rewards obtained as the training progresses. Aug 11, 2021 · Deep RL implementations. Lunar-Lander-Environment Code and relevant files for the final project of CM50270 (Reinforcement Learning) for MSc. It also introduces a target network which is updated periodically and helps the main network to reduce its loss easier. In our An implementation of Rainbow DQN on OpenAi's Lunar Lander Environment - kiranikram/Rainbow-DQN Neural network controlling a simulated spacecraft learning to safe land on a surface using reinforcement learning techniques. The challenge is to train a neural network to choose the best actions based on the state of the lander. DQN and variants SAC and PPO for the Lunar Lander Continuous environment by OpenAI - Rem4rkable/Solving-the-LunarLanderContinuous-v2 强化学习，在LunarLander任务环境上实现PPO算法、DQN算法以及DQN变体算法. Collection of my Reinforcement Learning (RL) practices including DQN, D3QN, and Adaptive Gamma, applied to the Lunar Lander and CartPole environments. The goal of this environment is to land the Lunar Lander on the Landing pad with zero speed. Dec 29, 2021 · DQN uses a neural network as a function estimator to estimate this Q-fuction, rather than storing the Q-values explicitely. LunarLander DQN PER Project LunarLander-v3 game played by an AI agent trained with the DQN algorithm with Prioritized Experience Replay (PER) using deep-reinforcement-learning python framework. To tackle this challenge, a Double Deep Q-Network (DDQN) (Hasselt et al. Contribute to midhoc/lunar_lander_dqn development by creating an account on GitHub. For the implementation of the actor-critic algorithm we loosely follow Ref. Contribute to svpino/lunar-lander development by creating an account on GitHub. DI-engine is a python library for solving general decision intelligence problems, which is based on implementations of reinforcement learning framework using PyTorch or JAX. With the power of PyTorch, it provides efficient training and evaluation of the agent. weinberg@mail. 2021 Fall Hack-A-Roo. This project uses Deep Q-Learning (DQN), a reinforcement learning algorithm where the agent This project implements two AI agents that learn to solve the Cart-Pole game. The goal of this Agent in the environment is to successfully land the Lunar-Lander with the correct speed and inclination from the observations it gets and his 4 possible actions. Refer to the report (LunarLander_Report. May 7, 2021 · In this post, We will take a hands-on-lab of Simple Deep Q-Network (DQN) on openAI LunarLander-v2 environment. utoronto. The agent learns through trial and error, using deep reinforcement learning to master the challenging task of lunar landing. Implements experience replay, target networks, and training visu This project implements Deep Q-Learning (DQN) and Double Deep Q-Learning (DDQN) to train an agent to solve the Lunar Lander problem, a classic Reinforcement Learning task from OpenAI's Gym. The goal is to train an agent to land a spacecraft safely between two flags using only thrust-based actions. An implementation of Rainbow DQN on OpenAi's Lunar Lander Environment - kiranikram/Rainbow-DQN In this project, we build a Dueling Deep Q Network with Experience Replay as an reinforcement learning agent to solve the discrete Lunar Lander environment in OpenAI Gym. Solving this Lunar Lander using traditional Tabular Methods is practically challenging and expensive due to its large state space and complex reward structure. Solving gym's lunar lander with deep Q network. - dqna64/lunar-lander-dqn Implementation of a DQN for the Lunar Lander env. By utilizing techniques such as experience replay and target networks, DQN effectively learns to solve complex environments like Gymnasium's Lunar Lander, demonstrating its potential in both gaming and real-world applications. A Deep Q-Network (DQN) implementation that trains an AI agent to successfully land a spacecraft using reinforcement learning. The agent utilizes a deep neural network architecture to approximate the Q-function, enabling it to make decisions based on the observed state of the environment. Includes neural network architectures, replay buffer, target networks, epsilon-greedy exploration, training loop, and performance evaluation. Focused on the LunarLander-v2 environment, the project features a simplified A Deep Reinforcement Learning Agent that learns how to Land a rocket from a Box2D Gymnasium Environment! - ghubnerr/lunar_lander This project implements deep Q learning (DQN) to optimize a lunar lander control policy. The project demonstrates reinforcement learning through experience replay and epsilon-greedy exploration. xfupnsbgddnesyrrtktiwpkdopfxvlfmucecjgnppawrdcgqwancogugmmsiivtmmzvhwwapmvk