Computer Science > Robotics
[Submitted on 20 Jun 2024]
Title:Adaptive Manipulation using Behavior Trees
View PDF HTML (experimental)Abstract:Many manipulation tasks use instances of a set of common motions, such as a twisting motion for tightening or loosening a valve. However, different instances of the same motion often require different environmental parameters (e.g. force/torque level), and thus different manipulation strategies to successfully complete; for example, grasping a valve handle from the side rather than head-on to increase applied torque. Humans can intuitively adapt their manipulation strategy to best suit such problems, but representing and implementing such behaviors for robots remains an open question. We present a behavior tree-based approach for adaptive manipulation, wherein the robot can reactively select from and switch between a discrete set of manipulation strategies during task execution. Furthermore, our approach allows the robot to learn from past attempts to optimize performance, for example learning the optimal strategy for different task instances. Our approach also allows the robot to preempt task failure and either change to a more feasible strategy or safely exit the task before catastrophic failure occurs. We propose a simple behavior tree design for general adaptive robot behavior and apply it in the context of industrial manipulation. The adaptive behavior outperformed all baseline behaviors that only used a single manipulation strategy, markedly reducing the number of attempts and overall time taken to complete the example tasks. Our results demonstrate potential for improved robustness and efficiency in task completion, reducing dependency on human supervision and intervention.
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.