European Space Agency
 

Welcome to SPARTAN Project

SPAring Robotics Technologies for Autonomous Navigation


The exploration of Mars is one of the main goals for both NASA and ESA, as confirmed by past and recent activities as well as future plans. Many of the key questions in solar system science might be addressed in an effective manner at Mars: the search for life, the understanding of the planets’ evolution, the solar system history, or even the inventorying of useful resources for future human exploration. The fact that Mars is also the more accessible and most Earth-like planet of our solar system facilitates addressing these goals, making the red planet the space exploration favourite target.

The last 15 years have set the sequence for the exploration of Mars as follows: 1) to identify from orbit interesting scientific and landing sites, 2) to explore/search for water on the ground, and 3) to investigate about possible human habitability conditions. Both on-orbit and surface missions have achieved remarkable results. While multiple and valuable investigations can be made at the surface of Mars, there is a clear consensus within the scientific community that the major scientific objectives of Martian exploration can only be achieved with the return of a sample to Earth.

Bringing Martian samples back to Earth would have the clear consequence of allowing intensive, different and detailed analysis of the collected Mars samples, even years after the return of the sample. The MSR scenario, as discussed at international level between NASA, ESA, CSA and JAXA within the iMARS would include two flight elements: an Orbiter and a Lander. The Orbiter and the Lander, launched separately to Mars, would work together to return at least a single Mars sample container back to Earth. After entering the Martian atmosphere, the Lander platform, featuring both a Sample Fetching Rover (SFR) and Mars Ascent Vehicle (MAV), would perform a soft landing on the Martian surface. The SFR will collect samples from the surface/subsurface, or pick up cached samples from a previous mission and return those back to the MAV. In both scenarios, emphasis is given to a reasonable mobility of such a rover, which must be at least in the range of future precision landing ellipse dimensions (< 10km) in the case of the SFR collecting cached samples or even up to 20 km in the scenario where the SFR will have to do the sampling.

In line with the above reported consideration and the requirement posed by ESA, the objectives of the SPARTAN activity is to reduce as much as possible the overall budgets required by the rover navigation function while improving on its performances (i.e. accuracy of terrain reconstruction, probability to find paths) so to make the system compatible with the requirements of a long traverse range capability device.

For this purpose, throughout the SPARTAN project, we aim to optimize the hardware implementation of the following sub-systems:

  • Imaging, implementing suitable local image processing that can serve Image products,
  • Visual Odometry which provides an estimation of the Displacement of the rover,
  • Visual SLAM which determines the current Location of the rover,
  • 3D Map reconstruction, which reconstructs the 3Dimensional shape of the terrain being imaged in front of the rover,
  • Localization.
More specifically, the SPARTAN project will be focused in the tight and optimal implementation of the computer vision algorithms for rover navigation using custom-designed vectorial processing (by means of FPGAs).
  • To identify recent computer vision algorithms suitable for robotic navigation (specifically 3D map reconstruction and localisation) that have the potential for implementation into parallel processing chains while having a high performance (i.e. accuracy and frame rate).
  • To trade off the algorithms w.r.t. potential for realisation in space-rated devices.
  • To implement the algorithms in a demonstration setup.
  • To test and demonstrate the implemented algorithms.
After this, the activity will focus first on a literature survey of computer vision algorithms regarding 3D reconstruction and Visual SLAM in order to produce a complete vision catalogue for rover navigation purposes. From the whole list of algorithms during the activity we will perform a trade-off based on criteria intended to minimize processor resources use such as time and memory and profit from the parallel processing inherent to FPGA technology.