Atari Asterix Environment

Overview

Asterix (Taz) was released by Atari in 1983 for the Atari 2600 and features the Looney Tunes character the Tasmanian Devil in a food frenzy. Within the game, Taz only appears as a tornado. The same game was released outside the United States featuring Asterix instead of Taz.

The gameplay is rather simple. The player guides Taz between the stage lines in order to eat hamburgers and avoid the dynamites. The game does not use any buttons and the difficulty increases by increasing the speed of the objects on screen. As the game progresses, the burgers may change into other edible or drinkable objects such as beer kegs, hot dogs, etc. There are not many sound effects in the game except a blipping sound when the player hits an edible object and another sound that resembles of explosion when the player hits dynamite.

Description from RetroGames

Performances of RL Agents

We list various reinforcement learning algorithms that were tested in this environment. These results are from RL Database. If this page was helpful, please consider giving a star!

Star

Human Starts

Result Algorithm Source
395599.5 Distributional DQN Rainbow: Combining Improvements in Deep Reinforcement Learning
364200.0 PDD DQN Dueling Network Architectures for Deep Reinforcement Learning
280114.0 Rainbow Rainbow: Combining Improvements in Deep Reinforcement Learning
31907.5 Prioritized DDQN (prop, tuned) Prioritized Experience Replay
22484.5 Prioritized DDQN (rank, tuned) Prioritized Experience Replay
22140.5 A3C FF Asynchronous Methods for Deep Reinforcement Learning
17244.5 A3C LSTM Asynchronous Methods for Deep Reinforcement Learning
16837.0 DDQN (tuned) Deep Reinforcement Learning with Double Q-learning
15840.0 DuDQN Dueling Network Architectures for Deep Reinforcement Learning
9199.5 Prioritized DQN (rank) Prioritized Experience Replay
7536.0 Human Massively Parallel Methods for Deep Reinforcement Learning
6723.0 A3C FF 1 day Asynchronous Methods for Deep Reinforcement Learning
5285.0 DDQN Deep Reinforcement Learning with Double Q-learning
3324.7 Gorila DQN Massively Parallel Methods for Deep Reinforcement Learning
164.5 Random Massively Parallel Methods for Deep Reinforcement Learning
124.5 DQN Massively Parallel Methods for Deep Reinforcement Learning

No-op Starts

Result Algorithm Source
454461 QR-DQN-0 Distributional Reinforcement Learning with Quantile Regression
428200.3 Rainbow Rainbow: Combining Improvements in Deep Reinforcement Learning
406211 C51 A Distributional Perspective on Reinforcement Learning
400529.5 Distributional DQN Rainbow: Combining Improvements in Deep Reinforcement Learning
375080.0 PDD DQN Dueling Network Architectures for Deep Reinforcement Learning
342016 IQN Implicit Quantile Networks for Distributional Reinforcement Learning
300732.0 IMPALA (deep) IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
261025 QR-DQN-1 Distributional Reinforcement Learning with Quantile Regression
205914.0 Reactor The Reactor: A fast and sample-efficient Actor-Critic agent for Reinforcement Learning
36238.5 Reactor The Reactor: A fast and sample-efficient Actor-Critic agent for Reinforcement Learning
32478 NoisyNet A3C Noisy Networks for Exploration
31583.0 ACKTR Scalable trust-region method for deep reinforcement learning using Kronecker-factored approximation
29692.5 IMPALA (shallow) IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
28350 NoisyNet DuDQN Noisy Networks for Exploration
28188.0 DuDQN Dueling Network Architectures for Deep Reinforcement Learning
17356.5 DDQN A Distributional Perspective on Reinforcement Learning
16121.0 Reactor ND The Reactor: A fast and sample-efficient Actor-Critic agent for Reinforcement Learning
15150.0 DDQN Deep Reinforcement Learning with Double Q-learning
14328 NoisyNet DQN Noisy Networks for Exploration
11170 DuDQN Noisy Networks for Exploration
8503.3 Human Dueling Network Architectures for Deep Reinforcement Learning
8503.3 Human Human-level control through deep reinforcement learning
6822 A3C Noisy Networks for Exploration
6433.33 Gorila DQN Massively Parallel Methods for Deep Reinforcement Learning
6253 DQN Noisy Networks for Exploration
6012 DQN Human-level control through deep reinforcement learning
4359.0 DQN A Distributional Perspective on Reinforcement Learning
2609.0 IMPALA (deep, multitask) IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
1332 Contingency Human-level control through deep reinforcement learning
987.3 Linear Human-level control through deep reinforcement learning
210.0 Random Human-level control through deep reinforcement learning

Normal Starts

Result Algorithm Source
6801.2 ACER Proximal Policy Optimization Algorithm
4532.5 PPO Proximal Policy Optimization Algorithm
3176.3 A2C Proximal Policy Optimization Algorithm