Probabilistic Safety Programs

TL;DR

This paper introduces Probabilistic Safety Programs (PSP), a framework that models environmental uncertainty and safety invariants to evaluate the safety of autonomous systems' future actions using graphical models and variational inference.

Contribution

The paper presents a novel probabilistic programming framework for safety evaluation that combines environmental uncertainty modeling with safety invariants, enabling efficient safety analysis.

Findings

Effective safety evaluation for quadrotors in dynamic environments.

Fast variational inference enables real-time safety assessment.

Framework generalizes to autonomous vehicles and other cyber-physical systems.

Abstract

Achieving safe control under uncertainty is a key problem that needs to be tackled for enabling real-world autonomous robots and cyber-physical systems. This paper introduces Probabilistic Safety Programs (PSP) that embed both the uncertainty in the environment as well as invariants that determine safety parameters. The goal of these PSPs is to evaluate future actions or trajectories and determine how likely it is that the system will stay safe under uncertainty. We propose to perform these evaluations by first compiling the PSP to a graphical model then using a fast variational inference algorithm. We highlight the efficacy of the framework on the task of safe control of quadrotors and autonomous vehicles in dynamic environments.