Agent and object detection and classification in a warehouse setting
|Title||Agent and object detection and classification in a warehouse setting|
|Summary||Detection, and classification of different agents (manual driven forklift trucks, other robots, humans) and objects (such as pallets) in a warehouse environment|
|Keywords||Robot perception, feature extraction, machine learning, classification|
|TimeFrame||October 2017 to June 2018, with possible extension to September 2018|
|References|| SAS2-project, http://islab.hh.se/mediawiki/SAS2
ROS - Robot Operating System, http://www.ros.org/
OpenCv - http://opencv.org/
Lalonde, Jean-Francois; Vandapel, Nicolas; Huber, Daniel; Hebert, Martial; Natural terrain classification using three-dimensional ladar data for ground robot mobility, Journal of Field Robotics, Vol. 23, No. 10, pp. 839 - 861, November, 2006
Mosberger, Rafael; Vision-based human detection from mobile machinery in industrial environments, Thesis, Örebro University, Sweden, 2016
Saarinen, Jari P.; Andreasson, Henrik; Stoyanov, Todor; Lilienthal, Achim J.; 3D normal distributions transform occupancy maps: An efficient representation for mapping in dynamic environments, The International Journal of Robotics Research, Vol 32, Issue 14, pp. 1627 – 1644, 2013
|Prerequisites||Programming skills (preferably C++ or Python)|
|Supervisor||Björn Åstrand, Naveed Muhammad|
A central issue for robots and automated guided vehicles (AGVs) is safety; the robot or AGV must not harm humans or damage objects in the environment. Safety concerns have become more and more important as the use of AGVs has spread and advances in sensor technology, sensor integration, and object detection and avoidance have been more widely adapted. Today’s safety systems don’t consider the identity of different agents in close proximity to the robot and AGV.
The goal of this project [as a subset of SAS2 project] is to develop a method for detection and identification of different agents and objects (such as other AGVs, manually driven forklift trucks, humans, pallets etc.) present in a warehouse environment. The idea is to investigate if a robot, using perception data it acquires through its exteroceptive sensors such as cameras and lidars, can detect and identify different categories of agents and objects present in its environment. This includes segmentation of perception data, extraction of features, and implementation of classification techniques. Preferably the solutions are designed as ROS-packages (or, c++, python, matlab-code).
Resources: Facilities for data logging, cameras, depth sensor, data logging equipment, data set from warehouses and collaboration with industrial partners.
Research Question: How to detect and classify different agents and objects present in a warehouse environment into different categories.
WP1: Literature review. WP2: Segmentation and feature extraction from perception data. WP3: Classification of different agents/objects using supervised and/or unsupervised learning methods. WP4: [bonus] conference publication (ICRA, IROS, ETFA, ECMR, TAROS)
Deliverable: an implementation and demonstration of the developed system for detection and classification of agents/objects using data from warehouse environments or mines.