Russian scientists have developed a prototype of a robotic bird featuring independent wings and tail feathers.

Birds in flight maneuver swiftly and accurately. Consequently, engineers aim to replicate the mechanics of their movements in artificial devices. To this end, Russian scientists have developed and constructed a robotic model of a feathered creature, which features independent control of each wing's movement, as well as adjustments for their span and the motion of the tail feathers.

The results of the research are published in the proceedings of the Sixth International Conference on "Neurotechnologies and Neurointerfaces." As the scientists explained, the distinctive feature of the proposed device is that each wing is equipped with a separate set of motors and control microelectronics. This results in the prototype having additional degrees of freedom that can replicate the motions of a real bird over a wide range. Thanks to this, the model mechanism allows for variations in the positions of the wings and tail feathers, helping to identify optimal modes for various maneuvers and types of flight.

“The prototype assists in comparing theoretical calculations with real conditions. With this model, it is possible to experimentally verify how factors such as wing positioning, flapping frequency, or the body position of the robotic bird in the airflow affect aerodynamics,” said research leader Viktor Kazantsev, head of the laboratory of neurobiomorphic technologies at MIPT and chair of the neurotechnology department at NNGU.

For instance, using the robotic model, scientists are determining the optimal angle for positioning the wing of a biomorphic drone relative to the oncoming airflow to enhance lift. They are also examining how to arrange the upper limbs relative to the torso to either accelerate or decelerate the robotic bird.

This data is crucial for designing control systems for artificial devices. In the future, it will be utilized for the initial training of neural networks, which will form the basis of the "brain" for the robotic birds.

“When creating nature-like mechanical devices, we inevitably deal with a large number of parameters that influence flight. The challenge of rapidly selecting optimal settings for successful maneuvering of the robotic bird in the airflow can be addressed using neural networks. Such software algorithms are currently being developed,” explained Viktor Kazantsev.

According to the scientists, as new settings are added to the robot during the machine learning process, the neural network will become more complex. Additionally, various sensors and technical vision devices will be integrated, enhancing the system's informational capacity.

Ultimately, the researchers aim to develop an onboard processor equipped with artificial intelligence technologies, which will be installed in production robotic birds.