Flying bugs are able to navigate their environments efficiently, processing visual stimuli to avert obstacles and land safely on a variety of surfaces. Over the past decade or so, research groups worldwide have been trying to replicate these capabilities in autonomous micro air vehicles (MAVs) utilizing mechanisms similar to these observed in bugs.
Researchers at TU Delft’s MAVLab have been trying to develop insect-inspired strategies that could improve navigation and landing methods in tiny drones for a number of years now. In a recent paper pre-published on arXiv, they launched a new technique for the creation of neuromorphic controllers that might improve landings in MAVs.
In previous work, researchers at the MAVLab created a series of bio-inspired methods for vision-based motion estimation utilizing spiking neural networks (SNNs). SNNs are a category of artificial neural networks (ANN) that closely mimic neural networks in the human mind, utilizing activation spikes to compute and analyze information.
In their new research, the Hagenaars and his colleagues wanted to take their strategies one step ahead, utilizing them to control the flight and landing of MAVs. To do that, they launched a partnership with the Dutch National research institute for computer science and mathematics, which has a high degree of expertise within the development of spiking neural networks.
Most previously developed methods to control MAVs throughout vision-based landing are based on proportional controllers and standard ANNs. Controllers powered by SNNs have the potential of achieving similar and even better outcomes with far greater energy efficiencies.