Combining object modeling and recognition for active scene exploration

Abstract

Active scene exploration incorporates object recognition methods for analyzing a scene of partially known objects and exploration approaches for autonomous modeling of unknown parts. In this work, recognition, exploration, and planning methods are extended and combined in a single scene exploration system, enabling advanced techniques such as multi-view recognition from planned view positions and iterative recognition by integration of new objects from a scene. Here, a geometry based approach is used for recognition, i.e. matching objects from a database. Unknown objects are autonomously modeled and added to the recognition database. Next-Best-View planning is performed both for recognition and modeling. Moreover, 3D measurements are merged in a Probabilistic Voxel Space, which is utilized for planning collision free paths, minimal occlusion views, and verifying the poses of the recognized objects against all previous information. Experiments on an industrial robot with attached 3D sensors are shown for scenes with household and industrial objects.

On-the-fly particle filter registration for laser data

“…During laser scans, the robot pose and range data are synchronized. Note that the depth measurements of the laser striper are very accurate in contrast to RGB-D cameras [2].…”

“…Therefore, many current registration methods are designed for data generated by 3D sensors such as Kinect. However, laser stripers are more accurate and can deal better with shiny or black objects, which is required for high quality 3D-modeling [2]. Although most of the existing registration methods work on laser data as well, they suffer from the time consuming data acquisition process of laser scanners, as they do not exploit the fact that partial data arrives continuously, i.e.…”

Skewness map: estimating object orientation for high speed 3D object retrieval system

“…In digital image processing, databases have become a common utility to be used. Image reconstruction (Kato and Ohta, 2014), object recognition (Kriegel et al, 2013), face recognition (Guo et al, 2010), and many other applications are some examples of a branch in image processing which are benefitted by the availability of databases. This also holds true to 3D modelling.…”

Using simplified geometric models in skill-based manipulation for objects used in daily life

“…[13,14] For example, there are studies on object modeling for grasping, object modeling that takes into account environmental knowledge, and object modeling for assemblies. [15][16][17] However, there are no studies which deal at the same time with object modeling for both planning and sensing in manipulation. [18] Figure 1.…”