Guided cost learning: deep inverse optimal control via policy optimization

Chelsea Finn, Sergey Levine, Pieter Abbeel

International Conference on Machine Learning · 2016 · 333 citations · 23 references

Concepts

TL;DR

Reinforcement learning can acquire complex behaviors from high‑level specifications, yet designing an effective cost function that encodes the correct task remains difficult. The study investigates using inverse optimal control to learn torque‑control behaviors from demonstrations for high‑dimensional robotic systems. The authors propose an algorithm that learns arbitrary nonlinear cost functions via neural networks without hand‑crafted features and introduces a sample‑based MaxEnt IOC approximation to handle unknown dynamics. Experiments on simulated and real robotic manipulation tasks show the method outperforms prior approaches in both task complexity and sample efficiency.

Abstract

Reinforcement learning can acquire complex behaviors from high-level specifications. However, defining a cost function that can be optimized effectively and encodes the correct task is challenging in practice. We explore how inverse optimal control (IOC) can be used to learn behaviors from demonstrations, with applications to torque control of high-dimensional robotic systems. Our method addresses two key challenges in inverse optimal control: first, the need for informative features and effective regularization to impose structure on the cost, and second, the difficulty of learning the cost function under unknown dynamics for high-dimensional continuous systems. To address the former challenge, we present an algorithm capable of learning arbitrary nonlinear cost functions, such as neural networks, without meticulous feature engineering. To address the latter challenge, we formulate an efficient sample-based approximation for MaxEnt IOC. We evaluate our method on a series of simulated tasks and real-world robotic manipulation problems, demonstrating substantial improvement over prior methods both in terms of task complexity and sample efficiency.

References

23