Learning Actions and Control of Focus of Attention with a Log-Polar-like Sensor
Robin Göransson, Volker Krueger
2023-09-22Atari Games
Abstract
With the long-term goal of reducing the image processing time on an autonomous mobile robot in mind we explore in this paper the use of log-polar like image data with gaze control. The gaze control is not done on the Cartesian image but on the log-polar like image data. For this we start out from the classic deep reinforcement learning approach for Atari games. We extend an A3C deep RL approach with an LSTM network, and we learn the policy for playing three Atari games and a policy for gaze control. While the Atari games already use low-resolution images of 80 by 80 pixels, we are able to further reduce the amount of image pixels by a factor of 5 without losing any gaming performance.
Related Papers
Generalized Adaptive Transfer Network: Enhancing Transfer Learning in Reinforcement Learning Across Domains2025-07-02A Principled Path to Fitted Distributional Evaluation2025-06-24Adaptive Action Duration with Contextual Bandits for Deep Reinforcement Learning in Dynamic Environments2025-06-17Meta-learning how to Share Credit among Macro-Actions2025-06-16TextAtari: 100K Frames Game Playing with Language Agents2025-06-04Improving Performance of Spike-based Deep Q-Learning using Ternary Neurons2025-06-03Automatic Reward Shaping from Confounded Offline Data2025-05-16Unraveling the Rainbow: can value-based methods schedule?2025-05-06