SEXTANT: Spartan EXTension Activity
Funding: European Space Agency (ESA)
Project Summary:
The SEXTANT activity builds upon the system developed within the SPARTAN (SPAring Robotics Technologies for Autonomous Navigation) project. SPARTAN targets at developing the hardware logic on FPGA of computer vision algorithms used by Martian rovers for visual navigation and localisation. The SPARTAN activity has already completed a first phase of computer-vision algorithms selection with a high potential for an efficient implementation fulfilling the application requirements. The SEXTANT objectives are intended to extend the breadth of the SPARTAN system to include: Implementation of two different set of computer vision algorithms used by Martian rovers in visual navigation and localisation. In order to fulfill this objective the Consortium is already revisiting the list of candidates proposed in SPARTAN as the trade-off for some criteria was based on estimates (the factual data can only be derived from the implementation) and it exists the possibility of having these runner-up algorithms having a performance similar or better than the initial selection made in SPARTAN and addressing the implementation of two different set of algorithms. It is also intended to perform a cross comparison in terms of performance with respect to different terrain typologies (sandy vs. rocky). Assessment of feasibility and prototyping of integrating orbital imagery to the Simultaneous Localisation and Mapping (SLAM) component of the system. In SPARTAN (and also in the first goal of SEXTANT, where two different sets of algorithms are to be implemented) the SLAM module solely relies on the visual features seen through the rover cameras. However, SLAM algorithms can work on maps where other unique features (e.g. boulders scattered on a plain) are evident. Therefore, the usage of prior topological information (such as high-resolution orbital maps of the Martian surface) to improve the absolute localisation accuracy of the rover will be addressed within the context of the SEXTANT activity. Although the SPARTAN system was already designed in a modular way that allows replacing its main components with different versions implementing the same functionality, some changes to its architecture are also foreseen to accommodate to extra needs imposed by the new algorithms.
Project Site: http://proteas.physics.auth.gr/sextant/
Consortium:
GMV AEROSPACE AND DEFENCE S.A.u. NATIONAL TECHNICAL UNIVERSITY OF ATHENS Foundation for Research and Technology - Hellas (FORTH)