Interpretable Control by Reinforcement Learning

TL;DR

This paper compares recent reinforcement learning methods that generate human-interpretable control policies for cart-pole balancing, demonstrating their effectiveness and real-world applicability through experiments and hardware implementation.

Contribution

It introduces and evaluates novel RL methods capable of producing human-interpretable policies as fuzzy controllers and algebraic equations, applicable to real-world systems.

Findings

Novel RL methods produce interpretable, high-performance policies

Methods successfully applied to hardware cart-pole system

Interpretable policies comparable to classical controllers

Abstract

In this paper, three recently introduced reinforcement learning (RL) methods are used to generate human-interpretable policies for the cart-pole balancing benchmark. The novel RL methods learn human-interpretable policies in the form of compact fuzzy controllers and simple algebraic equations. The representations as well as the achieved control performances are compared with two classical controller design methods and three non-interpretable RL methods. All eight methods utilize the same previously generated data batch and produce their controller offline - without interaction with the real benchmark dynamics. The experiments show that the novel RL methods are able to automatically generate well-performing policies which are at the same time human-interpretable. Furthermore, one of the methods is applied to automatically learn an equation-based policy for a hardware cart-pole demonstrator by using only human-player-generated batch data. The solution generated in the first attempt already represents a successful balancing policy, which demonstrates the methods applicability to real-world problems.