Addressing Myopic Constrained POMDP Planning with Recursive Dual Ascent
This work addresses suboptimal decision-making in CPOMDPs for applications like robotics or autonomous systems, but it appears incremental as it builds on existing Lagrangian-guided methods.
The paper tackled the problem of myopic action selection in constrained partially observable Markov decision processes (CPOMDPs) caused by global dual parameters in Lagrangian-guided Monte Carlo tree search, and introduced history-dependent dual variables with recursive dual ascent to improve exploration and safety outcomes.
Lagrangian-guided Monte Carlo tree search with global dual ascent has been applied to solve large constrained partially observable Markov decision processes (CPOMDPs) online. In this work, we demonstrate that these global dual parameters can lead to myopic action selection during exploration, ultimately leading to suboptimal decision making. To address this, we introduce history-dependent dual variables that guide local action selection and are optimized with recursive dual ascent. We empirically compare the performance of our approach on a motivating toy example and two large CPOMDPs, demonstrating improved exploration, and ultimately, safer outcomes.