ConstrainedZero: Chance-Constrained POMDP Planning using Learned Probabilistic Failure Surrogates and Adaptive Safety Constraints

TL;DR

This paper presents ConstrainedZero, a novel planning algorithm for CC-POMDPs that learns neural approximations of value, policy, and failure probability, enabling safe decision-making in uncertain environments with adaptive safety constraints.

Contribution

It introduces ConstrainedZero, combining neural network approximations with adaptive conformal inference for safe POMDP planning, improving safety without reward-cost trade-offs.

Findings

Achieves target safety levels in benchmark tasks.

Effectively balances safety and utility in uncertain environments.

Outperforms heuristic methods in safety-critical scenarios.

Abstract

To plan safely in uncertain environments, agents must balance utility with safety constraints. Safe planning problems can be modeled as a chance-constrained partially observable Markov decision process (CC-POMDP) and solutions often use expensive rollouts or heuristics to estimate the optimal value and action-selection policy. This work introduces the ConstrainedZero policy iteration algorithm that solves CC-POMDPs in belief space by learning neural network approximations of the optimal value and policy with an additional network head that estimates the failure probability given a belief. This failure probability guides safe action selection during online Monte Carlo tree search (MCTS). To avoid overemphasizing search based on the failure estimates, we introduce $\Delta$-MCTS, which uses adaptive conformal inference to update the failure threshold during planning. The approach is tested on a safety-critical POMDP benchmark, an aircraft collision avoidance system, and the sustainability problem of safe CO$_2$ storage. Results show that by separating safety constraints from the objective we can achieve a target level of safety without optimizing the balance between rewards and costs.