Effective and Robust Non-Prehensile Manipulation via Persistent Homology Guided Monte-Carlo Tree Search

2025-05-03 0 0 4.17MB 10 页 10玖币

侵权投诉

Eﬀective and Robust Non-Prehensile

Manipulation via Persistent Homology Guided

Monte-Carlo Tree Search

Ewerton R. Vieira1, Kai Gao2, Daniel Nakhimovich2,

Kostas E. Bekris2, and Jingjin Yu2⋆

1Dept. of Mathematics, Rutgers, NJ, USA

2Dept. of Computer Science, Rutgers, NJ, USA

Abstract. Performing object retrieval in real-world workspaces must

tackle challenges including uncertainty and clutter. One option is to apply

prehensile operations, which can be time consuming in highly-cluttered

scenarios. On the other hand, non-prehensile actions, such as pushing

simultaneously multiple objects, can help to quickly clear a cluttered

workspace and retrieve a target object. Such actions, however, can also

lead to increased uncertainty as it is diﬃcult to estimate the outcome

of pushing operations. The proposed framework in this work integrates

topological tools and Monte-Carlo Tree Search (MCTS) to achieve ef-

fective and robust pushing for object retrieval. It employs persistent

homology to automatically identify manageable clusters of blocking ob-

jects without the need for manually adjusting hyper-parameters. Then,

MCTS uses this information to explore feasible actions to push groups

of objects, aiming to minimize the number of operations needed to clear

the path to the target. Real-world experiments using a Baxter robot,

which involves some noise in actuation, show that the proposed frame-

work achieves a higher success rate in solving retrieval tasks in dense

clutter than alternatives. Moreover, it produces solutions with few push-

ing actions improving the overall execution time. More critically, it is

robust enough that it allows one to plan the sequence of actions oﬄine

and then execute them reliably on a Baxter robot.

Introduction video: youtu.be/00kEztqytRU .

Code: github.com/DanManN/planning_baxter/

Keywords: Manipulation, Topology, Monte-Carlo Tree Search.

1 Introduction

Retrieving a target object from a messy and constrained space, such as taking

out a bottle of water from a fridge, requires a robotic arm to relocate other

objects blocking access. Humans routinely perform such tasks with a high degree

of success, and the manipulation primitives used to execute such tasks are not

limited by pick-and-place-based rearrangements. Instead, they often involve non-

prehensile manipulation, such as pushing and pulling actions. Endowing robots

⋆This work supported by NSF CCF-2309866. EV is partially supported by Air Force

Oﬃce of Scientiﬁc Research under award numbers FA9550-23-1-0011 and FA9550-

23-1-0400.

arXiv:2210.01283v3 [cs.RO] 6 Feb 2024

2 Vieira et al.

with such skills is highly desirable, especially if they are tasked to carry out

ordinary human tasks. Humans execute such manipulation operations, including

grouping objects before performing a push, in a naturally robust manner, while

keeping the number of actions low.

Fig. 1. Top left: the experimental setup.

Top right: Scene S2 is one of the initial con-

ﬁgurations for the experiments. The robot

must move cluttered objects blocking access

to a target object Tplaced at the bottom

close to the wall. (2) to (5) demonstrate the

pushing actions executed after a given plan

by the proposed method. A simple grasping

plan is performed from (6) to (7).

This work proposes strategies for solv-

ing object retrieval problems focusing

on eﬃciency and robustness. An ex-

ample run is shown in Fig. 2, in which

the robot must retrieve the target ob-

ject located at the lower portion of the

workspace (a Pringles can of a speciﬁc

ﬂavor). Our study tackles the chal-

lenge by exploring two related sub-

strategies: (i) clustering objects to in-

form the choice of pushing actions,

and (ii) searching through and evalu-

ating feasible pushing operations be-

fore executing them in the real world.

The ﬁrst is motivated by human be-

haviors of implicitly grouping objects

before pushing them simultaneously.

Topological tools, such as persistent

homology, have been successfully em-

ployed to comprehensively identify

manageable object clusters for select-

ing the proper pushing actions [19].

The second direction leverages Monte-

Carlo Tree Search (MCTS) to explore the feasible identiﬁed pushing actions with

a high reward level in terms of the number of obstacles removed from the path

to the target and how dispersed the clusters are after the performed actions. The

reward metric is intuitive since a higher number of obstacles removed results in a

higher chance to solve the task faster, and dispersing the clusters leads to easier

pushing actions for consecutive steps.

The proposed framework integrates MCTS and the informed actions/rewards

provided by Persistent Homology. It is referred to here as PHIM (Persistent

Homology Informed actions and rewards for Monte-Carlo tree search). For real-

world experiments, the pipeline obtains the initial conﬁguration from perception

and then planning via PHIM to produce a high-level sequence of operations that

result in successful execution reliably. The solution is executed on the Baxter

robot. This can be done either by performing the plan open-loop or optionally re-

planning after each pushing action execution. Given the experimental evaluation,

PHIM achieves a signiﬁcantly higher success rate in solving cluttered problems

in constrained, shelf-like workspaces than alternatives, due to its long-horizon

planning capability and solution robustness. Furthermore, plans produced by

PHIM demonstrate are visually more natural and human-like.

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

EffectiveandRobustNon-PrehensileManipulationviaPersistentHomologyGuidedMonte-CarloTreeSearchEwertonR.Vieira1,KaiGao2,DanielNakhimovich2,KostasE.Bekris2,andJingjinYu2⋆1Dept.ofMathematics,Rutgers,NJ,USA2Dept.ofComputerScience,Rutgers,NJ,USAAbstract.Performingobjectretrievalinreal-worldworkspacesmustta...

展开>> 收起<<

Effective and Robust Non-Prehensile Manipulation via Persistent Homology Guided Monte-Carlo Tree Search.pdf

共10页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Effective and Robust Non-Prehensile Manipulation via Persistent Homology Guided Monte-Carlo Tree Search

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: