KABouM: Knowledge-Level Action and Bounding Geometry Motion Planner

Abstract

For robots to solve real world tasks, they often require the ability to reason about both symbolic and geometric knowledge. We present a framework, called KABouM, for integrating knowledge-level task planning and motion planning in a bounding geometry. By representing symbolic information at the knowledge level, we can model incomplete information, sensing actions and information gain; by representing all geometric entitiesobjects, robots and swept volumes of motions-by sets of convex polyhedra, we can efficiently plan manipulation actions and raise reasoning about geometric predicates, such as collisions, to the symbolic level. At the geometric level, we take advantage of our bounded convex decomposition and swept volume computation with quadratic convergence, and fast collision detection of convex bodies. We evaluate our approach on a wide set of problems using real robots, including tasks with multiple manipulators, sensing and branched plans, and mobile manipulation.

Planning with Perspectives -- Decomposing Epistemic Planning using Functional STRIPS

“…While in Cooper et al (2016)'s world, the seeing operator applies to propositional variables, and thus visibility can be interpreted more abstractly; for example, "seeing" (hearing) a message over a telephone. This connection between seeing and knowing is similar to the idea of sensing actions in partially-observable planning (Dornhege et al, 2009c;Dornhege, Eyerich, Keller, Trüg, Brenner, & Nebel, 2009a;Gaschler et al, 2018;Kaelbling & Lozano-Pérez, 2012;Bajada et al, 2015;Le et al, 2018;Cooper, Herzig, Maris, Perrotin, & Vianey, 2020;Kominis & Geffner, 2015, 2017Fabiano, Burigana, Dovier, & Pontelli, 2020), as seeing/sensing generates new knowledge. However, sensing actions are actions, whereas the idea of 'seeing' is a relation over properties of states.…”

“…They apply this idea by integrating with existing heuristic search-based planners. Their approach is widely used for robotic motion planning (Dornhege, Gissler, Teschner, & Nebel, 2009c;Gaschler, Petrick, Khatib, & Knoll, 2018;Kaelbling & Lozano-Pérez, 2012;Bajada, Fox, & Long, 2015). Planning Modulo Theories were introduced by Gregory, Long, Fox, and Beck (2012), an idea inspired by SAT Modulo Theories (Nieuwenhuis, Oliveras, & Tinelli, 2006), where specialized theories were integrated too with a heuristic search planner.…”

Hybrid conditional planning for robotic applications

“…Hybrid (classical) planning has been concerned about the problem of combining classical task planning with motion planning (Akbari et al, 2019; Beetz et al, 2012; Caldiran et al, 2009; Cambon et al, 2009; Colledanchise et al, 2016; Erdem et al, 2011; Gravot et al, 2003, 2006; Kaelbling and Lozano-Pérez, 2011; Lagriffoul et al, 2018; Lallement, 2016; Stock, 2017). Recent work on hybrid planning can be classified into three groups with respect to their computational approaches: (i) by developing/modifying search algorithms for task planning that utilize motion planning (Akbari et al, 2019; Beetz et al, 2012; Gravot et al, 2003; Hauser and Latombe, 2009; Kaelbling and Lozano-Pérez, 2013; Lagriffoul et al, 2014; Srivastava et al, 2014), (ii) by utilizing formal methods and relevant solvers (Dantam et al, 2016, 2018; Plaku, 2012), or (iii) by formally embedding motion planning as part of representations of actions and using relevant automated reasoners (Caldiran et al, 2009; Erdem et al, 2011, 2016; Gaschler et al, 2013, 2018; Hertle et al, 2012).…”