Deterministic versus Probabilistic Methods for Searching for an Evasive Target

Abstract

Several advanced applications of autonomous aerial vehicles in civilian and military contexts involve a searching agent with imperfect sensors that seeks to locate a mobile target in a given region. Effectively managing uncertainty is key to solving the related search problem, which is why all methods devised so far hinge on a probabilistic formulation of the problem and solve it through branch-and-bound algorithms, Bayesian filtering or POMDP solvers. In this paper, we consider a class of hard search tasks involving a target that exhibits an intentional evasive behaviour and moves over a large geographical area, i.e., a target that is particularly difficult to track down and uncertain to locate. We show that, even for such a complex problem, it is advantageous to compile its probabilistic structure into a deterministic model and use standard deterministic solvers to find solutions. In particular, we formulate the search problem for our uncooperative target both as a deterministic automated planning task and as a constraint programming task and show that in both cases our solution outperforms POMDPs methods.

Boosting Search Guidance in Problems with Semantic Attachments

“…In a Search-and-Track (SaT) mission (Stone 1975), a searching vehicle, the observer, wishes to locate the position of a moving object, the target, and to track it to a destination upon finding it. Bernardini et al (2016) propose a solution to SaT based on automated planning in which standard search patterns (e.g. spirals and lawnmowers) are employed by the observer to survey the search region.…”

“…In each plot, we show the results found in three cases: (i) without using the external advisor (no Adv) (Bernardini, Fox, and Long 2015); (ii) using the external advisor and a static approximation of the action effects on the total probabilities for heuristic guidance, i.e. approximation (iv) as described in Section 4.2 (Adv h (iv) ) (Bernardini et al 2016); and (iii) using the external advisor and exploiting it to dynamically calculate the action approximated effects on the total probabilities at each step for heuristic guidance, i.e. approximation (v) as described in Section 4.2 (Adv h (v) ).…”

“…This planner supports required concurrency, metric variables, predictable exogenous events and external advisors. The combination of a powerful planner with good heuristic guidance, both causal and numeric, allows us to successfully solve, within the same framework, complex problems in four very diverse domains: (i) Earth Observations problem (Aldinger and Löhr 2013); (ii) the Hydraulic Blocks World problem (Ivankovic et al 2014); (iii) the Voltage Control problem (Piacentini et al 2015); and (iv) large scale Search-and-Tracking missions (Bernardini et al 2016). When possible, we compare our solution to related techniques, providing evidence of the benefits of our approach.…”