Page - 39 - in Proceedings of the OAGM&ARW Joint Workshop - Vision, Automation and Robotics

AnAutonomous Transportation Robot for Urban Environments Konstantin Lassnig1 and Clemens Mu¨hlbacher2 and Gerald Steinbauer2 and Stefan Gspandl3 and Michael Reip3 Abstract—The transportation of goods is a central task of today’s economy. The cheap transportation of goods allows the wide spread of today’s internet based sales. To perform such transportation tasks one currently relies on humans. This imposes constraintswhen the transportation can be performed and imposes constraints on the costs. To address this time and cost constraints an automatic transportation of goods is preferred. Such an automatic transportation can be performed by an autonomousrobot, as theonesused inwarehouseenvironments. Although such environments are diverse andundergo a certain amount of change they are still rather static environments. To allow robots to perform the transportation in outdoor environments severalproblemsneed tobe tackled.Oneneeds to deal with large operation areas, uneven ground, and dynamic objects. In this paper, we present a robot system which can copewith these problems andallows to perform transportation tasks in outdoor environments. The focus of this paper will be on the localization andnavigation of the robotic in the outdoor environment allowing the robot to performoutdoor deliveries. I. INTRODUCTION The cheap transportation of goods is a central part of today’s economy. Reasonable prices of goods which are sold over the internet, heavily depend on transportation costs. Today’s supply chain requires a very dense distribution network and relies on the fact that sending a lot of packages on the same route is cheap. The larger number of goods for one route the cheaper it becomes. This is in contrast with the need for transporting goods to a single customer. Such a transportation is characterized by a few goods for one transportation route. To address this, robots offer a possible solution. Using a robot, the transportation can be performed in a flexible manner. Additionally, if multiple robots are used one can simply balance the load of transportation tasks on several robots. The use of a robot fleet for transportation tasks is getting adopted for warehouse environments nowadays [1], [2], [3]. These robot systems allow transporting goods in the ware- housewithout theneedofan adaptionof the warehouse.This is achieved by using algorithms allowing a localization and navigation in an indoor environment [4]. These algorithms use a 2D map of the environment. Such a map can be stored easily in the robots memory for a warehouse but not for 1Konstantin Lassnig is with ARTI, Graz Austria. This author was with the Institute for Software Technology when contributing, Graz University of Technology, Graz, Austria.klassnig@arti-robots.com 2Clemens Mu¨hlbacher and Gerald Steinbauer are with the Institute for Software Technology, Graz University of Technology, Graz, Austria. {cmuehlba,steinbauer}@ist.tugraz.at 3Stephan Gspandl and Michael Reip are with incubedIT, Hart bei Graz, Austria. {gspandl,reip}@incubedit.com large outdoor environments such as a city. Furthermore, the 2D map can be easily used in a warehouse for navigation as onecanassumea reasonableflatground.Suchanassumption cannot be made for an outdoor environment where the robot needs to ensure that it is not falling over road curbs. To allow a robot system to be used for transportation tasks in a large scale outdoor environment, one needs to address the problems which are imposed by the scale of the environment as well as the uneven ground. In this paper, we show a robot system which addresses these problems. The size of the environment is addressed by splitting the environment into smaller areas allowing the robot to keep only a small map in its memory. To allow the robot to be globally localized one additionally stores how the small pieces are related to each other. To tackle the uneven ground only the area close to the robot needs to be considered. This space is represented as a 2.5D surface and interpreted to find possible holes. The remainder of the paper is organized as follows. In the next section, we will discuss the software system used by the robot to perform transportation tasks in an outdoor environment.Theproceedingsectiondiscusseshowthe robot localizes itself despite the size of the environment. In Section IV we discuss how the robot navigates in the environment. This section also comprises a description how the robot deals with the uneven ground. Afterward, we discuss some related research. Finally, we conclude the paper and point out some future work. II. SYSTEM OVERVIEW Fig. 1. The transport robot [5]. In this paper, we discuss a robot which can perform a transportation task on a university campus autonomously. The robot can navigate indoor as well as outdoor. Further- more, the robot considers the uneven ground outdoors to 39