A Comparative Analysis of Interactive Reinforcement Learning Algorithms in Warehouse Robot Grid Based Environment

TL;DR

This paper compares two interactive reinforcement learning algorithms, Q-learning and SARSA, in a simulated warehouse environment to evaluate their effectiveness in training robots for complex tasks.

Contribution

It provides a systematic comparison of Q-learning and SARSA algorithms in a warehouse robot simulation, highlighting their relative performance and learning behaviors.

Findings

Q-learning outperformed SARSA in convergence speed.

Both algorithms successfully learned warehouse navigation tasks.

Consistent human feedback was crucial for fair comparison.

Abstract

The field of warehouse robotics is currently in high demand, with major technology and logistics companies making significant investments in these advanced systems. Training robots to operate in such complex environments is challenging, often requiring human supervision for adaptation and learning. Interactive reinforcement learning (IRL) is a key training methodology in human-computer interaction. This paper presents a comparative study of two IRL algorithms: Q-learning and SARSA, both trained in a virtual grid-simulation-based warehouse environment. To maintain consistent feedback rewards and avoid bias, feedback was provided by the same individual throughout the study.