In-Hand Pose Refinement Based on Contact Point Information

Iñigo Iturrate*, Yitaek Kim, Aljaz Kramberger, Christoffer Sloth

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review


This paper presents a method for in-hand pose refinement using point pair features and robust estimation. The algorithm assumes that an initial pose distribution is available, and uses this for an initial guess of the correspondences between source and target data. The data needed for the algorithm is points and surface normal vectors that can be estimated from force-torque data; thus, tactile sensor arrays are not needed. The method is demonstrated on the in-hand pose estimation of an object in an articulated five finger hand.

Original languageEnglish
Title of host publicationAdvances in Service and Industrial Robotics - RAAD 2023
EditorsTadej Petrič, Aleš Ude, Leon Žlajpah
PublisherSpringer Science+Business Media
Publication date2023
ISBN (Print)9783031326059
Publication statusPublished - 2023
Event32nd International Conference on Robotics in Alpe-Adria-Danube Region, RAAD 2023 - Bled, Slovenia
Duration: 14. Jun 202316. Jun 2023


Conference32nd International Conference on Robotics in Alpe-Adria-Danube Region, RAAD 2023
SeriesMechanisms and Machine Science
Volume135 MMS


  • in-hand pose estimation
  • robust optimization
  • tactile manipulation


Dive into the research topics of 'In-Hand Pose Refinement Based on Contact Point Information'. Together they form a unique fingerprint.

Cite this