Limits of Probabilistic Safety Guarantees when Considering Human Uncertainty
This work highlights a critical limitation in ensuring safety for autonomous systems interacting with humans, pointing out that existing approaches are inadequate for real-world safety standards.
The paper argues that for safety-critical applications like autonomous driving, probabilistic safety guarantees require extremely low failure probabilities (δ < 10^-8), but current learning-based methods fail to compute accurate confidence bounds at such levels due to inaccurate distributional assumptions and infeasible data requirements, as demonstrated using human driving and synthetic data.
When autonomous robots interact with humans, such as during autonomous driving, explicit safety guarantees are crucial in order to avoid potentially life-threatening accidents. Many data-driven methods have explored learning probabilistic bounds over human agents' trajectories (i.e. confidence tubes that contain trajectories with probability $δ$), which can then be used to guarantee safety with probability $1-δ$. However, almost all existing works consider $δ\geq 0.001$. The purpose of this paper is to argue that (1) in safety-critical applications, it is necessary to provide safety guarantees with $δ< 10^{-8}$, and (2) current learning-based methods are ill-equipped to compute accurate confidence bounds at such low $δ$. Using human driving data (from the highD dataset), as well as synthetically generated data, we show that current uncertainty models use inaccurate distributional assumptions to describe human behavior and/or require infeasible amounts of data to accurately learn confidence bounds for $δ\leq 10^{-8}$. These two issues result in unreliable confidence bounds, which can have dangerous implications if deployed on safety-critical systems.