Danger recognition and visual aid for the monitoring of wayfinding robots in obstacle avoidance scenarios
Authors
Schmuck, Viktor ; Polack, Sonny Glasius
Term
4. term
Education
Publication year
2018
Submitted on
2018-05-28
Pages
48
Abstract
Fremskridt inden for undvigelse af forhindringer over de sidste 15 år har åbnet nye muligheder for indendørs navigation i trange, menneskefyldte områder. Alligevel er det stadig udfordrende at opfatte omgivelserne korrekt og bevæge sig sikkert. Samtidig bliver robotter en større del af hverdagen, hvilket øger behovet for at forstå, hvordan de interagerer med mennesker og omgivelser. Der er fortsat et hul i opgavestyring og i at gøre robotters "hjerne"—deres beslutningsprocesser—synlig og forståelig. Vi præsenterer en løsning, der kan genkende farlige scenarier, samt et endnu ikke afprøvet visuelt værktøj, som hjælper med at forstå, hvordan en robot reagerer på input. Vi konkluderer, at vores løsning er programmatisk korrekt, men kræver forbedringer, før den kan anvendes i virkelige situationer.
Advances in obstacle avoidance over the past 15 years have opened new possibilities for indoor navigation in crowded spaces. Yet it remains challenging to perceive the environment and move safely. As robots become part of everyday life, there is a growing need to understand how they interact with people and their surroundings. Key gaps persist in task management and in making a robot’s “brain”—its decision processes—visible and understandable. We present a solution that recognizes dangerous scenarios and an untested visual tool to help explain how a robot reacts to inputs. We conclude that our solution is programmatically accurate but needs improvements before it can be used in real-world settings.
[This abstract was generated with the help of AI]
Keywords
Documents