Enhancing Sampling-based Planning with a Library of Paths

TL;DR

This work tackles efficient path planning for solid 3D objects moving in environments with narrow passages, where the state is in $SE(3)$. It introduces RRT-LIB, which builds a library of template paths via RRT-IR and reuses them for new, similar objects by identifying the closest template with a 3D shape similarity method, aligning paths with ICP, and guiding the planner along the transformed paths. The approach yields up to an $85\%$ reduction in planning time and high success rates in challenging scenarios, outperforming standard OMPL planners and solving cases they cannot. By enabling cross-object transfer of planning knowledge in repeated environments, RRT-LIB offers practical benefits for tasks like bin-picking, and an open-source implementation is provided for the community.

Abstract

Path planning for 3D solid objects is a challenging problem, requiring a search in a six-dimensional configuration space, which is, nevertheless, essential in many robotic applications such as bin-picking and assembly. The commonly used sampling-based planners, such as Rapidly-exploring Random Trees, struggle with narrow passages where the sampling probability is low, increasing the time needed to find a solution. In scenarios like robotic bin-picking, various objects must be transported through the same environment. However, traditional planners start from scratch each time, losing valuable information gained during the planning process. We address this by using a library of past solutions, allowing the reuse of previous experiences even when planning for a new, previously unseen object. Paths for a set of objects are stored, and when planning for a new object, we find the most similar one in the library and use its paths as approximate solutions, adjusting for possible mutual transformations. The configuration space is then sampled along the approximate paths. Our method is tested in various narrow passage scenarios and compared with state-of-the-art methods from the OMPL library. Results show significant speed improvements (up to 85% decrease in the required time) of our method, often finding a solution in cases where the other planners fail. Our implementation of the proposed method is released as an open-source package.

Figures (12)

• Figure 1: RRT-LIB: Rapidly-exploring Random Trees with a Library of Paths --- a novel algorithm for path planning in environments with narrow passages, consisting of the preparation phase (top) and the planning phase (bottom).
• Figure 2: The preparation phase of the proposed RRT-LIB algorithm. The planner iteratively finds multiple paths through the environment. Using the already-found paths as input to the planner increases the probability of finding distinct paths. After enough distinct paths are found, the planning is terminated, and the paths are saved in the library.
• Figure 3: The planning phase of the proposed RRT-LIB algorithm. The input consists of a manipulated object and an environment. Based on the inputs, paths computed for the most similar object in the library are loaded. After transforming the paths to account for possibly different positions of the manipulated object and the object from the library, we gain paths hinting at the possible paths through the environment. Increasing the sampling rate along these guiding paths should increase the probability of finding a solution.
• Figure 4: For every query object (left on each image), the most similar object from the library is selected (right on each image). The correspondences between the objects (illustrated by the colored lines) are used as the initial guess in the ICP algorithm.
• Figure 5: When the mutual position and rotation of the guiding object (red) and the similar object (blue) are not taken into account, the guiding paths become useless because they guide the similar object through the wall instead of the middle window (\ref{['fig:guiding_transformed_a']}). After the guiding paths are transformed using the transformation obtained from ICP, planning along them becomes viable (\ref{['fig:guiding_transformed_b']}).