Practical Local Planning in the Contact Space

Abstract

Proximity query is an integral part of any motion planning algorithm and takes up the majority of planning time. Due to performance issues, most existing planners perform queries at fixed sampled configurations, sometimes resulting in missed collisions. Moreover, randomly determining collision-free configurations makes it difficult to obtain samples close to, or on, the surface of C-obstacles in the configuration space. In this paper, we present an efficient and practical local planning method in contact space which uses "continuous collision detection" (CCD). We show how, using the precise contact information provided by a CCD algorithm, a randomized planner can be enhanced by efficiently sampling the contact space, as well as by constraining the sampling when the roadmap is expanded. We have included our contact-space planning methods in a freely available stateof-the-art planning library-the Stanford MPK library. We have been able to observe that in complex scenarios involving cluttered and narrow passages, which are typically difficult for randomized planners, the enhanced planner offers up to 70 times performance improvement when our contact-space sampling and constrained sampling methods are enabled.

Retraction-based RRT planner for articulated models

“…The retraction-based approaches have been widely used to improve the performance of sample-based planners in narrow passages [1], [21], [23], [28], [31]. The main idea is to retract a randomly generated configuration that lies in CObstacle space towards a more desirable region, e.g.…”

“…Other methods perform contact space planning, i.e. generate more samples that touch the boundaries of C-Obstacles [21], [22]. Based on efficient penetration depth computation, Zhang et al [31] present a retraction-based planner for rigid models.…”

A Fast and Practical Algorithm for Generalized Penetration Depth Computation

“…A few good algorithms are known for convex polytopes [19], [20] and general polygonal models [8]. Due to the difficulty of computing a global PD t between non-convex models, some local PD t algorithms have been proposed [9], [10], [21].…”

“…For each basic contact constraint C i , we compute its Jacobian, which is the normal of the corresponding parameterized configuration space. Using this normal, we obtain a half-plane, which is a linearization of the contact surface [21], [37]. By concatenating the halfplanes using Boolean operators • i , we generate a non-convex polyhedral cone, which serves as a local linear approximation of C contact .…”

C²A: Controlled conservative advancement for continuous collision detection of polygonal models

“…Thus, CCD never misses any potential collision between configurations. CCD algorithms are used in sampling-based motion planning to perform the local planning step [3], [4], [5], and in robot dynamics to find the first time of contact to apply responsive forces [6]. However, the major drawback of CCD algorithms are that they are typically slower than the discrete counterpart.…”