Probabilistic Plan Recognition Using Off-the-Shelf Classical Planners

By Miquel Ramírez, Hector Geffner. 2010

379 citations recorded by the source corpus.

Plan recognition is the problem of inferring the goals and plans of an agent after observing its behavior. Recently, it has been shown that this problem can be solved efficiently, without the need of a plan library, using slightly modified planning algorithms. In this work, we extend this approach to the more general problem of probabilistic plan recognition where a probability distribution over the set of goals is sought under the assumptions that actions have deterministic effects and both agent and observer have complete information about the initial state. We show that this problem can be solved efficiently using classical planners provided that the probability of a partially observed execution given a goal is defined in terms of the cost difference of achieving the goal under two conditions: complying with the observations, and not complying with them. This cost, and hence the posterior goal probabilities, are computed by means of two calls to a classical planner that no longer has to be modified in any way. A number of examples is considered to illustrate the quality, flexibility, and scalability of the approach.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Plan explicability and predictability for robot task planning (2015, background): “There exists a rich literature on plan recognition [15], [4], [21], [17], and many recent works use these techniques in human-aware planning”
Deliberation for autonomous robots: A survey (2017, background): “The latter rely on deterministic [116, 182] or probabilistic [87] planning techniques, as well as parsing techniques [177].”
Combining Planning with Gaze for Online Human Intention Recognition (2018, methodology): “Such models use belief-desire-intention plans [26] or planning models [24, 29].”
A review of generalized planning (2019, methodology): “This approach is useful for object classification [55] but also for goal recognition [76] and task classification [86].”
Model Recognition as Planning (2019, methodology): “Once the planning model of the observed agent is recognized, the modelbased machinery for automated planning becomes fully applicable for other recognition tasks like goal recognition (Ramı́rez and Geffner 2010; Ramı́rez 2012), goal recognition design (Keren, Gal, and Karpas 2014) or counterplanning”
Inventing Relational State and Action Abstractions for Effective and Efficient Bilevel Planning (2022, background): “Finally, inverse planning [36, 38, 85] and inverse reinforcement learning [35, 86, 87, 88] study the problem of leveraging demonstrations to infer cost functions and, in turn, policies.”
RADAR - A Proactive Decision Support System for Human-in-the-Loop Planning (2017, methodology): “Plan Correction or Repair In the event πh is invalid and may be repaired with additional actions, we can leverage the compilation pr2plan from (Ramı́rez and Geffner 2010) for a slightly different outcome.”
Automated Planning and Acting (2016, methodology): “Among these are Hector Geffner, Robert Goldman, Patrik Haslum, Jörg Hoffmann, Joachim Hertzberg, Felix Ingrand, Ugur Kuter, Marco Pistore, Mak Roberts, Vikas Shivashankar, Sylvie Thiébaux, and Qiang Yang.”
Autonomous agents modelling other agents: A comprehensive survey and open problems (2017, methodology): “(Ramı́rez and Geffner, 2010) yes no yes yes no –∗∗ discrete full”
Coordination in human-robot teams using mental modeling and plan recognition (2014, methodology): “For the plan recognition component, we used the probabilistic plan recognition algorithm developed by Ramirez and Geffner [9].”