Decentralized drone swarms deployed today either rely on sharing of positions among agents or detecting swarm members with the help of visual markers. This work proposes an entirely visual approach to coordinate markerless drone swarms based on imitation learning. Each agent is controlled by a small and efficient convolutional neural network that takes raw omnidirectional images as inputs and predicts 3D velocity commands that match those computed by a flocking algorithm. We start training in simulation and propose a simple yet effective unsupervised domain adaptation approach to transfer the learned controller to the real world. We further train the controller with data collected in our motion capture hall. We show that the convolutional neural network trained on the visual inputs of the drone can learn not only robust inter-agent collision avoidance but also cohesion of the swarm in a sample-efficient manner. The neural controller effectively learns to localize other agents in the visual input, which we show by visualizing the regions with the most influence on the motion of an agent. We remove the dependence on sharing positions among swarm members by taking only local visual information into account for control. Our work can therefore be seen as the first step towards a fully decentralized, vision-based swarm without the need for communication or visual markers.

Flight through cluttered environments, such as forests, poses great challenges for animals and machines alike because even small changes in flight path may lead to collisions with nearby obstacles. When flying along narrow corridors, insects use the magnitude of visual motion experienced in each eye to control their position, height, and speed but it is unclear how this strategy would work when the environment contains nearby obstacles against a distant background. To minimise the risk of collisions, we would expect animals to rely on the visual motion generated by only the nearby obstacles but is this the case? To answer this, we combine behavioural experiments with numerical simulations and provide the first evidence that bumblebees extract the maximum rate of image motion in the frontal visual field to steer away from obstacles. Our findings also suggest that bumblebees use different optic flow calculations to control lateral position, speed, and height.

Abstract-Most existing robots are designed to exploit only one single locomotion mode, such as rolling, walking, flying, swimming, or jumping, which limits their flexibility and adaptability to different environments where specific and different locomotion capabilities could be more effective. Here we introduce the concept and the design of a flying robot with Adaptive Morphology for Multi-Modal Locomotion. We present a prototype that can use its wings to walk on the ground and fly forward. The wings are used as whegs to move on rough terrains. This solution allows to minimize the structural mass of the robot by reusing the same structure (here the wings) for different modes of locomotion. Furthermore, the morphology of the robot is analysed and optimized for ground speed.

Elementary Motion Detectors (EMD) are well-established models of visual motion estimation in insects. The response of EMDs are tuned to specific temporal and spatial frequencies of the input stimuli, which matches the behavioural response of insects to wide-field image rotation, called the optomotor response. However, other behaviours, such as speed and position control, cannot be fully accounted for by EMDs because these behaviours are largely unaffected by image properties and appear to be controlled by the ratio between the flight speed and the distance to an object, defined here as relative nearness. We present a method that resolves this inconsistency by extracting an unambiguous estimate of relative nearness from the output of an EMD array. Our method is suitable for estimation of relative nearness in planar scenes such as when flying above the ground or beside large flat objects. We demonstrate closed loop control of the lateral position and forward velocity of a simulated agent flying in a corridor. This finding may explain how insects can measure relative nearness and control their flight despite the frequency tuning of EMDs. Our method also provides engineers with a relative nearness estimation technique that benefits from the low computational cost of EMDs.

No abstract