Johann Prankl

About

Johann Prankl is a leading researcher in robotic perception and 3D computer vision, whose work focuses on enabling robots to understand and interact with complex, unstructured environments. His major contributions span object segmentation, 3D reconstruction, and semantic scene understanding, with a strong emphasis on practical applications for service and industrial robotics. Prankl’s most influential work, “Segmentation of unknown objects in indoor environments” (156 citations), introduced a robust framework for handling clutter and occlusion in RGB-D images, directly supporting tasks like object search and manipulation. He further advanced pose estimation with his multi-task template matching approach (49 citations), addressing the challenge of occluded, texture-less objects. His innovative “3-D Entangled Forests” method (45 citations) improved semantic segmentation of point clouds by learning contextual relationships between structures. Notably, Prankl also developed ScalableFusion (19 citations), a high-resolution, real-time 3D reconstruction system that enhances color fidelity beyond sensor limitations. Beyond these technical achievements, his work on unsupervised dirt spot detection for floor cleaning robots (23 citations) and calibration of vignetting effects (14 citations) demonstrates a commitment to solving real-world robotics challenges. With over 370 total citations, Prankl’s research continues to shape how robots perceive, model, and navigate their surroundings.

Research Focus

Key Achievements

Top Papers

1. 1
   Segmentation of unknown objects in indoor environments

   156 citations · 2012
2. 2
   Multi-Task Template Matching for Object Detection, Segmentation and Pose Estimation Using Depth Images

   49 citations · 2019
3. 3
   Enhancing Semantic Segmentation for Robotics: The Power of 3-D Entangled Forests

   45 citations · 2015
4. 4
   RGB-D object modelling for object recognition and tracking

   41 citations · 2015
5. 5
   The Power of GMMs: Unsupervised Dirt Spot Detection for Industrial Floor Cleaning Robots

   23 citations · 2017
6. 6
   ScalableFusion: High-resolution Mesh-based Real-time 3D Reconstruction

   19 citations · 2019
7. 7
   Knowing your limits - self-evaluation and prediction in object recognition

   14 citations · 2011
8. 8
   Calibration and correction of vignetting effects with an application to 3D mapping

   14 citations · 2016
9. 9
   Interactive object modelling based on piecewise planar surface patches

   7 citations · 2013
10. 10
    Learning the Floor Type for Automated Detection of Dirt Spots for Robotic Floor Cleaning Using Gaussian Mixture Models

    6 citations · 2017

Key Collaborators

Contact & Links