Learning Actions and Control of Focus of Attention with a Log-Polar-like Sensor

TL;DR

This paper demonstrates that using log-polar like image data with gaze control and deep reinforcement learning can significantly reduce image data size without sacrificing performance in Atari games.

Contribution

It introduces a novel approach combining log-polar image representation with deep RL for gaze control, reducing data size while maintaining performance.

Findings

Reduced image data by a factor of 5 without performance loss

Extended A3C with LSTM for gaze control in Atari games

Effective use of log-polar images for attention-focused processing

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.