Dr.-Ing. Sylvain Joyeux
Navigation algorithms for unknown or badly known terrains: using the robot body-sensing capabilities to improve navigation robustness.
|Duration:||01.04.2009 till 30.06.2012|
|Donee:||German Research Center for Artificial Intelligence GmbH|
Federal Ministry of Economics and Technology
German Aerospace Center e.V.
|Grant number:||This project is funded by the Space Agency of the German Aerospace Center with federal funds of the Federal Ministry of Economics and Technology (BMWi) in accordance with the parliamentary resolution of the German Parliament, grant no. 50RA0907|
SAR- & Security Robotics
Advanced Security Guard V3
In the Intelligent Mobility project, DFKI RIC Bremen deals with the four main issues in autonomous navigation: robot control, mapping, localization and decision-making.
One quite classical problem with current navigation approaches is that the robot can have illogical information about his underground, for example. The originality of this project is to use body-sensing capabilities (like touch, vibration, motor efforts) to improve navigation capabilities. Contrary to most approaches, DFKI RIC Bremen will solve this problem by fully integration of information the robot body provides for navigation. The system will therefore be able to detect physically impossible estimations.
DFKI RIC Bremen has already developed several autonomous robotic systems. One of them is the successfully proven and tested platform ASGUARD. The project team will first integrate all the necessary sensors on this platform. Ways to control ASGUARD optimally on various terrains types (rubber, rocks, concrete) and high slopes will be studied. Finally, the team will use its body-sensing capabilities for terrain recognition. Recognizing the terrain based on how it “feels”: hard or soft, rough or flat, … .
In a second part, the project will focus on the integration of body sensing in localization and mapping. The main problem here is that robots need to know their position well to build a good map and they needs to know the map well to estimate their position in it. To solve that cross-dependency, a specific class of algorithms already exists: the SLAM (Simultaneous Localization And Mapping) algorithms. This project will give birth to embodied SLAM (eSLAM) algorithms, where the SLAM approaches will be improved by using the information coming from the robot’s body
Finally, the DFKI RIC Bremen will study the supervision of autonomous navigation: the ability to detect problems and correct them before they become critical. This includes also the ability for the robot to make informed decisions about the paths it should take, as for instance deciding to cross a bridge instead of using ASGUARD’s swimming capabilities.
Initial tests in the field of the current version 4 of the ASGUARD system
IMoby: Simple speed control vs. wheel synchronisation
Simple Speed Control vs. Wheel Synchronization
ASGUARD v2: Intelligent mobility (Final promotion)
Final Promotion Video for Asguard v2 in the iMoby-projekt.