# Deep Reinforcement Learning for Industrial Insertion Tasks with Visual Inputs and Natural Rewards

## TL;DR

This paper demonstrates that combining reinforcement learning with prior information enables effective learning of industrial insertion tasks using visual inputs and natural rewards, overcoming challenges of sample efficiency and safety in real-world settings.

## Contribution

It introduces a method that integrates RL with classical controllers or demonstrations to solve complex industrial insertion tasks from limited real-world interactions.

## Key findings

- Successful learning of insertion tasks with visual inputs.
- Effective handling of sparse and goal-based rewards.
- Reduced sample complexity in real-world environments.

## Abstract

Connector insertion and many other tasks commonly found in modern manufacturing settings involve complex contact dynamics and friction. Since it is difficult to capture related physical effects with first-order modeling, traditional control methods often result in brittle and inaccurate controllers, which have to be manually tuned. Reinforcement learning (RL) methods have been demonstrated to be capable of learning controllers in such environments from autonomous interaction with the environment, but running RL algorithms in the real world poses sample efficiency and safety challenges. Moreover, in practical real-world settings we cannot assume access to perfect state information or dense reward signals. In this paper, we consider a variety of difficult industrial insertion tasks with visual inputs and different natural reward specifications, namely sparse rewards and goal images. We show that methods that combine RL with prior information, such as classical controllers or demonstrations, can solve these tasks from a reasonable amount of real-world interaction.