• Home
  • Search
  • ConstrainedZero: Chance-Constrained POMDP Planning Using Learned Probabilistic Failure Surrogates and Adaptive Safety Constraints
  • Open Access IconOpen Access
  • Cite Icon1
  • https://doi.org/10.24963/ijcai.2024/746Copy DOI Icon

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

  • Aug 1, 2024
  • Chao Xu +3 more
Show More
  • Abstract
  • Literature Map
  • Citations
  • Similar Papers
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 Δ-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 CO₂ 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.

Similar Papers
  • Conference Article
  • Citations3

Monte Carlo Tree Search with Variable Simulation Periods for Continuously Running Tasks

  • Nov 01, 2019
  • Seydou Ba +4
  • Research Article

Boosting MCTS With Free Energy Minimization.

  • Nov 18, 2025
  • Neural computation
  • Mawaba Pascal Dao +1
  • Conference Article

Environmental Map Learning with Multiple-Robots

  • May 19, 2025
  • Azin Shamshirgaran +1
  • Research Article
  • Citations8

Measurement Maximizing Adaptive Sampling with Risk Bounding Functions

  • Jul 17, 2019
  • Proceedings of the AAAI Conference on Artificial Intelligence
  • Benjamin Ayton +2
  • PDF
  • Research Article
  • Citations53

Adaptive Stress Testing: Finding Likely Failure Events with Reinforcement Learning

  • Dec 06, 2020
  • Journal of Artificial Intelligence Research
  • Ritchie Lee +7
  • Research Article

Threshold UCT: Cost-Constrained Monte Carlo Tree Search with Pareto Curves

  • Apr 11, 2025
  • Proceedings of the AAAI Conference on Artificial Intelligence
  • Martin Kurečka +3
  • Conference Article

Point-Based Incremental Pruning for Monte-Carlo Tree Search

  • Jul 01, 2017
  • Bo Wu +1
  • PDF
  • Research Article
  • Citations7

Scalable Online Planning for Multi-Agent MDPs

  • Mar 10, 2022
  • Journal of Artificial Intelligence Research
  • Shushman Choudhury +3
  • Research Article
  • Citations19

Simplified decision making in the belief space using belief sparsification

  • Apr 01, 2022
  • The International Journal of Robotics Research
  • Khen Elimelech +1
  • Research Article

Semi‐Markovian planning to coordinate aerial and maritime medical evacuation platforms

  • Aug 26, 2025
  • AI Magazine
  • Mahdi Al‐Husseini +2
  • Book Chapter
  • Citations8

Research on Motion Planning of Seven Degree of Freedom Manipulator Based on DDPG

  • Dec 15, 2018
  • Li-Lan Liu +3
  • Book Chapter
  • Citations2

An Improved Genetic Algorithm and Its Application in Constrained Solid TSP in Uncertain Environments

  • Jan 01, 2015
  • Monoranjan Maiti +2
  • PDF
  • Research Article
  • Citations268

Finding Approximate POMDP solutions Through Belief Compression

  • Jan 01, 2005
  • Journal of Artificial Intelligence Research
  • N Roy +2
  • Research Article
  • Citations21

The role of surface acoustic wave technology in communication systems

  • Mar 01, 1972
  • Ultrasonics
  • J.H Collins +1
  • Research Article
  • Citations746

Decision Making in Recurrent Neuronal Circuits

  • Oct 01, 2008
  • Neuron
  • Xiao-Jing Wang
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.