Robust Entry Descent and Landing System for ExoMars

FEIC feature tracking on a PANGU generated image sequence

The Robust Entry Descent and Landing System (EDLS) for ExoMars study aims to develop technologies to ensure a safe landing of ExoMars on the surface of Mars. The overall project is being led by LogicaCMG.

The overall project objectives are to:

  • Assess a number of novel but pragmatic and achievable candidate methodologies to:
    • Improve the accuracy of a landing from ballistic entry.
    • Reduce the risk of mission failure during the terminal phase.
  • Quantify the impact of key atmospheric variables on the propagation of the injection covariance matrix during entry, descent and landing thereby allowing:
    • A more rigorous approach to be taken to the design of the overall EDLS solution
    • Novel methods for accuracy improvement and risk reduction to be developed in a realistic environment

The principal objective for the University of Dundee was to:

  • Investigate processing of sensor data to estimate the transverse velocity of the lander.

Excessive transverse velocity, primarily due to wind, has to be reduced to below 15 m/s for a safe landing. To null the transverse velocity its magnitude must be known. The University of Dundee developed image processing algorithms to detect the transverse velocity of the lander. These were based on the Feature Extraction Integrated Circuit (FEIC) that University of Dundee developed for the ESA Navigation for Planetary Approach and Landing study led by EADS Astrium in France. This study developed an intelligent camera for vision-based navigation of a planetary lander. The FEIC chip in this camera performs image processing to select image feature points and to track them from frame to frame.

Measurement of the transverse velocity of the spacecraft was made using the FEIC algorithm to perform visual feature tracking. This information was then combined with altitude information to estimate the motion of the spacecraft. The algorithms were tested using the PANGU simulation tool. A known trajectory (series of spacecraft positions and orientations where each sensor measurement is made) was used to produce the image sequence. The images were processed and the transverse velocity estimated. This was then compared to the known transverse velocity from the trajectory.

Using PANGU, Monte-Carlo testing was carried out with various wind speeds and initial positions over a simulated Martian surface. Accurate velocity component calculation to within 0.5 m/s was demonstrated in 97% of cases.

Further work will involve improving the image-processing algorithms, and carrying out more realistic and demanding tests with more complex camera motion.