View source for Pallet Detection and Mapping

{{StudentProjectTemplate
|Summary=Pallet Detection and Mapping in a Warehouse Environment
|Programme=Mobile and Autonomous Systems
|Keywords=Object Recognition, Classification, Mapping, Scene Understanding
|TimeFrame=January 2015 until June 2015,  with possible extension until September 2015.
|References=Belongie, Serge, Jitendra Malik, and Jan Puzicha. "Shape matching and object recognition using shape contexts." Pattern Analysis and Machine Intelligence, IEEE Transactions on 24.4 (2002): 509-522.

Viola, Paul, and Michael Jones. "Rapid object detection using a boosted cascade of simple features." Computer Vision and Pattern Recognition, 2001. CVPR 2001. Proceedings of the 2001 IEEE Computer Society Conference on. Vol. 1. IEEE, 2001.

Lowe, David G. "Local feature view clustering for 3D object recognition." Computer Vision and Pattern Recognition, 2001. CVPR 2001. Proceedings of the 2001 IEEE Computer Society Conference on. Vol. 1. IEEE, 2001.

Lowe, David G. "Distinctive image features from scale-invariant keypoints." International journal of computer vision 60.2 (2004): 91-110.

Bay, Herbert, et al. ”Speeded-up robust features (SURF).” Computer vision and image understanding 110.3 (2008): 346-359.

Pinto, Nicolas, David D. Cox, and James J. DiCarlo. "Why is real-world visual object recognition hard?." PLoS computational biology 4.1 (2008): e27.
|Prerequisites=Image analysis, programming skills (preferably C++ or Python), familiarity with pattern recognition, Familiarity with filtering techniques (eg. EKF) for mobile robots mapping.
|Supervisor=Björn Åstrand, Saeed Gholami Shahbandi,
|Level=Master
|Status=Open
}}
concise description:
This project [as a subset of AIMS project], targets the automation of the logistic management system of a warehouse, by the means of automatic guided vehicles (AGVs, eg. lift trucks). Achieving this objective is feasible if the operating robots (AGVs) understand their surrounding through a high level model of the world. An essential element of this model is the inventory list. Locating stored articles and modulating the inventory list with respect to human’s expectation delivers a more effective system. A common element of the storage in warehouses is the pallet. By detecting and mapping the pallets in a warehouse, the problem of constructing the inventory list will be simplified to the problem of identification of those objects stored over each pallet.

RQ: Detection of pallets falls into the category of object recognition. Although the pallet’s pattern is relatively simple and unique, yet the recognition process will be challenging. To reach a reliable result, one should deal with problems such as segmenting stacked pallets, cluttered environment, bad illumination and different view points.

WP1: startup: literature review and data acquisition
WP2: pallet detection
WP3: pallet mapping 2D, [3D: bonus]

Deliverable: an implementation and demonstration of the developed method for detection and mapping pallets, based on a dataset acquired in a real world warehouse.
[bonus] conference publication (ETFA, ECMR, TAROS).