As part of our Bionic Learning Network, we have been exploring the wonder of flight for over 15 years. In that time, we have researched numerous flying objects and their natural principles, learning from these biological models to create our own technology based on them. Autonomous swarm behaviour still presented us with a big challenge. With the BionicBee, for the first time our team has now developed a flying object that can fly in large numbers and completely autonomously in a swarm.
At around 34 grams in weight, 22 centimetres in length and with a wingspan of 24 centimetres, the BionicBee is the smallest flying object from the Bionic Learning Network to date. This is the first time that the developers used the generative design method: after entering a few parameters, a software finds the optimal structure based on specific design principles in order to use as little material as necessary to create the most stable design possible. This consistent lightweight construction is crucial for good manoeuvrability and flight duration.
The bee’s body forms the compact housing for the beating wing mechanism, the communication technology as well as the control components for the wing beats and the adaptation of the wing geometry. A brushless motor, three servo motors, the battery, the gear unit and various PCBs are installed in the tightest of spaces. The intelligent interaction between the motors and the mechanics means that the frequency of the wing beats can, for example, be precisely adjusted for the various manoeuvres.
The artificial bee flies with a wing beat frequency of 15 to 20 hertz. The wings beat back and forth at a 180-degree angle. The brushless motor drives the wing beats without backlash via a precisely guided, ultra-light mechanical construction. The higher the speed, the higher the wing beat frequency and the lift. The three servo motors at the base of the wing change the geometry of the wing in a particular way, thus increasing the effectiveness of certain wing positions and generating a specific variation of the lift.
If the bee is supposed to fly forward, the geometry is adjusted so that the lift in the rear wing position is greater than in the forward position. This causes the body to tilt forward (pitch), and the bee flies forwards. If the geometry is adjusted so that the right wing generates more lift than the left wing, the bee rolls around the longitudinal axis to the left and flies off to the side. Another option is to adjust it in such a way that one wing generates more lift at the front and the second wing generates more lift at the rear. This causes the bee to rotate (yaw) around the vertical axis.
The autonomous behaviour of the ten bees is made possible by an indoor localisation system with ultra-wideband technology (UWB). Eight UWB anchors are installed on two levels in the space. This means that the runtime can be accurately measured and the bees can position themselves in the space. The UWB anchors send signals to the individual bees, which independently measure the distances to the transmitters and can calculate their own position in the space using the time stamps.
To fly in a swarm, the bees follow the paths specified by a central computer. High spatial and temporal accuracy is required for safe and collision-free flight in close formation. Possible mutual interplay caused by air turbulence ("downwash") must also be taken into account when planning the route.
As each bee is built by hand, even the smallest manufacturing differences can influence flight behaviour. That's why the bees also have an automatic calibration function, so that after a short test flight, each bee determines its own optimised controller parameters. This is how the intelligent algorithm can work out the hardware differences between the individual bees. And that, in turn, allows the entire swarm to be controlled externally as if all the bees were identical.
When developing the BionicBee, the developers took advantage of the numerous insights they gained during previous projects. It is the next in a series of bionic flying objects that have been created as part of our Bionic Learning Network. For over 15 years, we have been developing research platforms using general technical principles that are based on nature. Click on the links below to have a quick look at them.