Michele Gattullo 2 yr. doctoral program - XXVIII cycle DMMM·ING-IND/15·VR 3 Lab

 Fite-Georgel (2011): Is there a reality for IAR? ◦ 2 apps (54 total) out of the laboratory; ◦ 1 still in use; ◦ Main reasons:  scalability and reproducibility  scarce collaboration with industries  Research goal: find suitable solutions for ◦ scenario ◦ hardware ◦ software

 IAR solutions can be commercialized only when they are: ◦ Reliable: high accuracy, fall-back solutions testing in lab ◦ User friendly: safe and easy to set up, learn, use, and customize. user tests ◦ Scalable: easily reproducible and distributable in large numbers commercially available solutions

Common manual assembly and maintenance operations: spot welding, testing, packing, repairing, inspecting.  Reproducible in VR 3 Lab Reliable  Low/medium level tasks User friendly  Hardware/Software independent of product and operations Scalable

 Tracking and camera pose estimation  AR content management  User Interface navigation

 Marker based tracking: realiable but not scalable  3D SLAM extended tracking ◦ sparse 3D representation of the surrounding environment ◦ Initial image: control scale and placing  3D tracking based on CAD data SLAM CAD

Goal: Find the best (reliable, user friendly, scalable) tracking technique for AR. Workflow:  State of art  Algorithms study  Demo development and testing

 Tracking errors cannot be eliminated Tracking tolerant approach Maintenance/Assembly Operation Is the operation complex/unintuitive? 3D tracking techniques Is the tracking quality acceptable? 3D AR animation on Handeld devices, lightweight HWDs 3D animation (no AR) on HUDs 2D projected signs (Spatial AR) NO YES

Goal: Definition and validation of the «tracking tolerant approach». Workflow:  Maintenance/assembly tasks classification  Case studies creation and testing  ∀ task find how (2D signs, 3D signs, 3D anim) and where (projected, handheld, HUD, HWD) display information

 Voice recognition  Gesture recognition: Kinect, Leap Motion  Hand recognition: based on image processing

Goal: Find the best (reliable, user friendly, scalable) UI navigation method for AR. Workflow:  State of art  NUIs development  NUIs testing and comparison

Michele Gattullo Contact information: Dipartimento di Meccanica, Matematica e Management Viale Japigia 182, Bari, IT Terza palazzina (stanza prof. Monno) Tel skype: michele.gattullo