Flapping wing micro air vehicles (MAVs) take inspiration from natural fliers, such as insects and hummingbirds. Existing designs manage to mimic the wing motion of natural fliers to a certain extent; nevertheless, differences will always exist due to completely different building blocks of biological and man-made systems. The same holds true for the design of the wings themselves, as biological and engineering materials differ significantly. This paper presents results of experimental optimization of wing shape of a flexible wing for a hummingbird-sized flapping wing MAV. During the experiments we varied the wing 'slackness' (defined by a camber angle), the wing shape (determined by the aspect and taper ratios) and the surface area. Apart from the generated lift, we also evaluated the overall power efficiency of the flapping wing MAV achieved with the various wing design. The results indicate that especially the camber angle and aspect ratio have a critical impact on the force production and efficiency. The best performance was obtained with a wing of trapezoidal shape with a straight leading edge and an aspect ratio of 9.3, both parameters being very similar to a typical hummingbird wing. Finally, the wing performance was demonstrated by a lift-off of a 17.2 g flapping wing robot.

Hovering flapping flight is inherently unstable and needs to be stabilized actively. We present a control mechanism that modulates independently the wing flapping amplitude and offset by displacing joints of a flapping linkage mechanism. We demonstrate its performance by high speed camera recordings of the wing motion as well as by direct measurements of pitch moment and lift force. While flapping at 17 Hz the prototype produces 90 mN of lift and generates pitch moments from-0.7 N.mm to 1.1 N.mm. The mechanism shows low level of cross-coupling in combined pitch and roll commands.

Abstract-In this paper, we propose an efficient implementation of the branch and bound method for knapsack problems on a CPU-GPU system via CUDA. Branch and bound computations can be carried out either on the CPU or on a GPU according to the size of the branch and bound list. A better management of GPUs memories, less GPU-CPU communications and better synchronization between GPU threads are proposed in this new implementation in order to increase efficiency. Indeed, a series of computational results is displayed and analyzed showing a substantial speedup on a Tesla C2050 GPU.