Neural Manipulation Planning on Constraint Manifolds

The presence of task constraints imposes a significant challenge to motion planning. Despite all recent advancements, existing algorithms are still computationally expensive for most planning problems. In this letter, we present Constrained Motion Planning Networks (CoMPNet), the first neural planner for multimodal kinematic constraints. Our approach comprises the following components: i) constraint and environment perception encoders; ii) neural robot configuration generator that outputs configurations on/near the constraint manifold(s), and iii) a bidirectional planning algorithm that takes the generated configurations to create a feasible robot motion trajectory. We show that CoMPNet solves practical motion planning tasks involving both unconstrained and constrained problems. Furthermore, it generalizes to new unseen locations of the objects, i.e., not seen during training, in the given environments with high success rates. When compared to the state-of-the-art constrained motion planning algorithms, CoMPNet outperforms by order of magnitude improvement in computational speed with a significantly lower variance.

I. Introduction

Efficient and scalable manipulation planning is of paramount importance in robotics and automation to solve real-world tasks. However, in most cases, manipulation of objects imposes hard kinematic constraints on the robot that limit its allowable range of motion. Examples of such instances include moving an object from one place to another that might also contain orientation constraints [1], maintaining contact with the environment such as in robot surgery [2], and performing bi-manual manipulation [3]. In all of these cases, kinematic constraints form one or more low-dimensional manifolds of a set of robot configurations embedded in a high-dimensional ambient space, and are often known as manifold constraints [4].