Safe Reinforcement Learning Using Advantage-Based Intervention

06/16/2021
by   Nolan Wagener, et al.
0

Many sequential decision problems involve finding a policy that maximizes total reward while obeying safety constraints. Although much recent research has focused on the development of safe reinforcement learning (RL) algorithms that produce a safe policy after training, ensuring safety during training as well remains an open problem. A fundamental challenge is performing exploration while still satisfying constraints in an unknown Markov decision process (MDP). In this work, we address this problem for the chance-constrained setting. We propose a new algorithm, SAILR, that uses an intervention mechanism based on advantage functions to keep the agent safe throughout training and optimizes the agent's policy using off-the-shelf RL algorithms designed for unconstrained MDPs. Our method comes with strong guarantees on safety during both training and deployment (i.e., after training and without the intervention mechanism) and policy performance compared to the optimal safety-constrained policy. In our experiments, we show that SAILR violates constraints far less during training than standard safe RL and constrained MDP approaches and converges to a well-performing policy that can be deployed safely without intervention. Our code is available at https://github.com/nolanwagener/safe_rl.

READ FULL TEXT
research
02/14/2022

SAUTE RL: Almost Surely Safe Reinforcement Learning Using State Augmentation

Satisfying safety constraints almost surely (or with probability one) ca...
research
09/20/2019

Reconnaissance and Planning algorithm for constrained MDP

Practical reinforcement learning problems are often formulated as constr...
research
07/09/2021

Safe Exploration by Solving Early Terminated MDP

Safe exploration is crucial for the real-world application of reinforcem...
research
05/29/2022

On the Robustness of Safe Reinforcement Learning under Observational Perturbations

Safe reinforcement learning (RL) trains a policy to maximize the task re...
research
08/26/2020

Constrained Markov Decision Processes via Backward Value Functions

Although Reinforcement Learning (RL) algorithms have found tremendous su...
research
02/17/2022

Efficient Learning of Safe Driving Policy via Human-AI Copilot Optimization

Human intervention is an effective way to inject human knowledge into th...
research
06/07/2021

Closed-Form Analytical Results for Maximum Entropy Reinforcement Learning

We introduce a mapping between Maximum Entropy Reinforcement Learning (M...

Please sign up or login with your details

Forgot password? Click here to reset