Perching combines the best of both climbing and flying platforms. While flying, the vehicle is agile and can cover long distances quickly. When perched, the same vehicle now has reduced energy expenditure, is quiet and can wait for better flying conditions. Perching could be used in search-and-rescue or surveillance operations, and opens the way to many new applications including reconfigurable sensor networks, close proximity inspection of structure or sample-and-return missions.

In this thesis, we discuss an approach allowing small airplanes to land on rough vertical surfaces using micro-spines. The technique presented consists in flying toward the wall at cruise speed (10m/s), triggering a pitch up maneuver at the appropriate distance from the wall to rapidly slow down and absorb the remaining kinetic energy at impact with a suspension that also favors the engagements of the micro-spines. This suspension creates a landing envelope of successful touchdown conditions, which can be easily reached by the airplane assuming a properly designed airframe.

This thesis proposes and validates an hybrid model of the suspension used to determine the size of the landing envelope. Furthermore, to create the largest landing envelope possible, an extension to current region of attraction estimation techniques is proposed. This extension uses sums-of-squares optimization, which combines polynomial approximations of barrier constraints with the traditional Lyapunov methods to achieve tight estimation in the presence of barriers. Finally, the flying dynamics are analyzed and a simple model is developed by approximating the maneuver in three distinct phases based on the angle of attack of the wing (AoA).

The results presented provide a better understanding of the parameters determining the shape of the landing envelope created by the suspension, but also the parameters of the airframes shaping the funnel that brings the airplane from normal flying conditions to the landing envelope. Furthermore, the region of attraction extension was tested on a simple model of the airplane and showed a 12x improvement in the area size estimated. Ultimately, the airplane designed with the insights gained with that work was able to perform 20 consecutive autonomous landings in outdoor conditions using only onboard sensors (i.e. laser range finder, accelerometers and gyroscopes).