# Probabilistically Safe Policy Transfer

## TL;DR

This paper introduces a probabilistic framework for safe policy transfer in robotics, ensuring damage constraints are maintained during learning while optimizing performance.

## Contribution

It proposes a novel method that predicts safety impacts of policy updates, enabling safe learning with damage constraints at every step.

## Key findings

- The approach effectively balances performance improvement and safety constraints.
- Experiments demonstrate the method maintains safety while enhancing robot performance.
- The method predicts safety outcomes of policy modifications in real-time.

## Abstract

Although learning-based methods have great potential for robotics, one concern is that a robot that updates its parameters might cause large amounts of damage before it learns the optimal policy. We formalize the idea of safe learning in a probabilistic sense by defining an optimization problem: we desire to maximize the expected return while keeping the expected damage below a given safety limit. We study this optimization for the case of a robot manipulator with safety-based torque limits. We would like to ensure that the damage constraint is maintained at every step of the optimization and not just at convergence. To achieve this aim, we introduce a novel method which predicts how modifying the torque limit, as well as how updating the policy parameters, might affect the robot's safety. We show through a number of experiments that our approach allows the robot to improve its performance while ensuring that the expected damage constraint is not violated during the learning process.