Combining Learning from Human Feedback and Knowledge Engineering to Solve Hierarchical Tasks in Minecraft

TL;DR

This paper presents a hybrid approach combining imitation learning, human feedback, and knowledge engineering to solve complex hierarchical tasks in Minecraft, winning the 2021 NeurIPS MineRL BASALT Challenge.

Contribution

It introduces a novel hybrid method that integrates human demonstration data, feedback, and engineered modules for hierarchical task solving in a complex environment.

Findings

Achieved first place in the NeurIPS MineRL BASALT Challenge

Outperformed end-to-end machine learning approaches

Provided a human-like, interpretable agent solution

Abstract

Real-world tasks of interest are generally poorly defined by human-readable descriptions and have no pre-defined reward signals unless it is defined by a human designer. Conversely, data-driven algorithms are often designed to solve a specific, narrowly defined, task with performance metrics that drives the agent's learning. In this work, we present the solution that won first place and was awarded the most human-like agent in the 2021 NeurIPS Competition MineRL BASALT Challenge: Learning from Human Feedback in Minecraft, which challenged participants to use human data to solve four tasks defined only by a natural language description and no reward function. Our approach uses the available human demonstration data to train an imitation learning policy for navigation and additional human feedback to train an image classifier. These modules, combined with an estimated odometry map, become a powerful state-machine designed to utilize human knowledge in a natural hierarchical paradigm. We compare this hybrid intelligence approach to both end-to-end machine learning and pure engineered solutions, which are then judged by human evaluators. Codebase is available at https://github.com/viniciusguigo/kairos_minerl_basalt.