Suchergebnisse

Robotter vil om få år blive en del af vores daglige liv. Inden for produktionsindustrien har det allerede være tilfældet i mange år, men anvendelsen af mobile robotter har hidtil været henvist til isolerede områder som græsslåning, overvågning, landbrugsproduktion og militære funktioner. Fremskridt i forskningen gør, at robotter vil kunne assistere os i mange af vore daglige gøremål i en ikke så fjern fremtid. En af de teknologier, der skal gøre dette muligt, er navigation, og navigation er emnet for denne afhandling. Navigation for autonome robotter handler om robottens evne til autonomt at manøvrere fra den nuværende position til et ønsket bestemmelsessted. Denne afhandling præsenterer og validerer eksperimentelt løsninger til detektering af farbar vej, omgåelse af forhindringer og gennemførelse af missioner. Vejklassifikationen er baseret på laserscanner-målinger og assisteret med vision for længere rækkevidde. Vejklassifikationen er tilstrækkelig selektiv til at kunne adskille selv flade vejrabatter fra selve vejen og kan isolere asfaltveje fra grusveje. Vejgenkendelse ud fra kamerabilleder tager udgangspunkt i klassifikationen fra laserscanneren og bruger en kombination af farve og kantdetektering til at estimere farbar vej på længere afstande. Resultatet af disse to sensorer anvendes under planlægning af en farbar rute, og her er det især robottens evne til at følge vejens kanter, der er undersøgt. Navigationen i en mission er styret af et sekventielt manuskript. Manuskriptsproget giver mulighed for detektering af vejkryds, for beregninger til brug under passagen af disse kryds og til valg a styringsmetode i øvrigt. Det samlede system er testet på en kombination af asfalt- og grusveje, med et antal forgreninger og vejkryds. Missioner på op til 3km længde er gennemkørt autonomt med det beskrevne system. Fokus i afhandlingen har været den eksperimentelle validering af de implementerede metoder og metodernes evne til at løse problemer i en virkelig verden. Der skal en betydelig mængde software til for at styre en autonom robot, emner som software genbrug og distribueret udvikling er derfor essentielle. Denne afhandling beskriver yderligere en komponentbaseret arkitektur for robotter, som kan fremme software genbrug og distribueret udvikling og vedligeholdelse. ; Abstract Robots will soon take part in everyone's daily life. In industrial production this has been the case for many years, but up to now the use of mobile robots has been limited to a few and isolated applications like lawn mowing, surveillance, agricultural production and military applications. The research is now progressing towards autonomous robots which will be able to assist us in our daily life. One of the enabling technologies is navigation, and navigation is the subject of this thesis. Navigation of an autonomous robot is concerned with the ability of the robot to direct itself from the current position to a desired destination. This thesis presents and experimentally validates solutions for road classification, obstacle avoidance and mission execution. The road classification is based on laser scanner measurements and supported at longer ranges by vision. The road classification is sufficiently sensitive to separate the road from flat roadsides, and to distinguish asphalt roads from gravelled roads. The vision-based road detection uses a combination of chromaticity and edge detection to outline the traversable part of the road based on a laser scanner classified sample area. The perception of these two sensors are utilised by a path planner to allow a number of drive modes, and especially the ability to follow road edges are investigated. The navigation mission is controlled by a script language. The navigation script controls route sequencing, junction detection, junction crossing calculations and drive mode selection. The entire system is tested on a combination of narrow asphalt and gravelled roads connected by a number of junctions. Missions of up to 3km in length have been successfully completed using the described system. The main focus of the thesis has been the experimental validation of the implemented solutions and the ability of the methods to solve real world problems. The amount of software needed by an autonomous robot can be overwhelming. Software reuse and distributed development are therefore important issues. The thesis describes a new component architecture for robotics software that promotes software reuse and distributed development and maintenance.

"The extraordinary inside story of the LEGO company--producer of the most beloved and popular toy on the planet--based on unprecedented access to the founding family that still owns the company, chronicling the brand's improbable journey to become the empire that it is today"--

Many studies demonstrate differences in the coverage of citing publications in Google Scholar (GS) and Web of Science (WoS). Here, we examine to what extent citation data from the two databases reflect the scholarly impact of women and men differently. Our conjecture is that WoS carries an indirect gender bias in its selection criteria for citation sources that GS avoids due to criteria that are more inclusive. Using a sample of 1250 U.S. researchers in Sociology, Political Science, Economics, Cardiology and Chemistry, we examine gender differences in the average citation coverage of the two databases. We also calculate database-specific h-indices for all authors in the sample. In repeated simulations of hiring scenarios, we use these indices to examine whether women's appointment rates increase if hiring decisions rely on data from GS in lieu of WoS. We find no systematic gender differences in the citation coverage of the two databases. Further, our results indicate marginal to non-existing effects of database selection on women's success-rates in the simulations. In line with the existing literature, we find the citation coverage in WoS to be largest in Cardiology and Chemistry and smallest in Political Science and Sociology. The concordance between author-based h-indices measured by GS and WoS is largest for Chemistry followed by Cardiology, Political Science, Sociology and Economics.

High unemployment is a problem in most European countries. The starting point of this article is that structural problems in the labour market and economic disincentives due to benefit and tax systems are the main reason why Europe faces severe unemployment. Many suggestions have been made to solve the problem, but few countries have succeeded in fighting unemployment. This paper points out that Denmark has succeeded in fighting unemployment with a battery of instruments primarily structural in kind. Along the same lines, Europe also has to structure tax and benefit systems to make work pay. Especially high Net Replacement Rates at the lower end of the wage distribution create disincentives to work and reforms need to be implemented to make the benefit system more employment friendly. One suggestion - which this paper discusses - is to introduce an employment-conditional scheme through the tax system - the so-called Earned Income Tax Credit (EITC). The EITC improves employment opportunities for low skilled workers by strengthening incentives to work without reducing benefit entitlements. This is achieved by introducing a tax credit given to low income people with a job. Simulations suggest that the EITC can improve employment although it is not a 'free lunch'. Distortions arise from financing the EITC and from the withdrawal of the tax credit.