[Paper Review] Plan Development using Local Probabilistic Models

Ella Atkins, Edmund H. Durfee+1 more|arXiv (Cornell University)|Feb 13, 2013

Real-Time Systems SchedulingComputer Science13 references20 citations

TL;DR

This paper proposes a local probabilistic modeling approach to enhance plan development in dynamic environments by assigning time-dependent probability functions to state transitions. By computing local state probabilities, the method selects high-probability goal paths and prunes unlikely states, significantly improving performance in the CIRCA architecture as demonstrated in flight simulation tests.

ABSTRACT

Approximate models of world state transitions are necessary when building plans for complex systems operating in dynamic environments. External event probabilities can depend on state feature values as well as time spent in that particular state. We assign temporally -dependent probability functions to state transitions. These functions are used to locally compute state probabilities, which are then used to select highly probable goal paths and eliminate improbable states. This probabilistic model has been implemented in the Cooperative Intelligent Real-time Control Architecture (CIRCA), which combines an AI planner with a separate real-time system such that plans are developed, scheduled, and executed with real-time guarantees. We present flight simulation tests that demonstrate how our probabilistic model may improve CIRCA performance.

Motivation & Objective

• To address the challenge of planning in complex, dynamic environments where world state transitions are uncertain and depend on state features and time.
• To develop a method that models transition probabilities locally and temporally to improve plan reliability and efficiency.
• To integrate probabilistic state modeling into the CIRCA architecture for real-time plan development with guaranteed execution.
• To evaluate the impact of local probabilistic models on plan quality and system performance through simulation.

Proposed method

• The approach assigns temporally dependent probability functions to state transitions based on state feature values and time spent in each state.
• Local state probabilities are computed using these time-dependent functions to assess the likelihood of future states.
• High-probability goal paths are selected while low-probability states are pruned during plan generation.
• The model is embedded within the CIRCA architecture, which couples an AI planner with a real-time execution system to ensure plan feasibility.
• The probabilistic model enables dynamic adaptation of plans in response to changing environmental conditions.
• Flight simulations are used to evaluate the model’s impact on plan development performance.

Experimental results

Research questions

• RQ1 How can local probabilistic models improve plan selection in dynamic, uncertain environments?
• RQ2 To what extent do time-dependent transition probabilities enhance plan reliability and efficiency?
• RQ3 How does integrating local probabilistic modeling into CIRCA affect real-time plan execution performance?
• RQ4 What is the impact of pruning low-probability states on overall plan quality?

Key findings

• The integration of local probabilistic models into CIRCA significantly improves plan development performance in dynamic environments.
• The use of time-dependent probability functions enables more accurate estimation of state transition likelihoods.
• High-probability goal paths are consistently selected, while improbable states are effectively pruned during planning.
• Flight simulation results demonstrate measurable improvements in plan quality and system responsiveness.