Espen Oland Control System Manager
Kitemill AS Evangerveien 3 5704 Voss Norway eo@kitemill.no www.kitemill.com
A Solution to the Pose Estimation Problem for Airborne Wind
Energy Systems using Multiple Bluetooth 5.1 Devices
Espen Oland Kitemill AS
The common approach to estimate the position and the attitude of a rigid body outdoors is to use an inertial mea-surement unit (IMU) providing meamea-surements of the an-gular velocity and linear acceleration vector and combine it with a magnetometer to provide attitude estimates. The position is commonly measured using GPS signals and combining the IMU and GPS through sensor fusion leads to very good estimates of the pose of a rigid body. Accelerometers estimate the roll and pitch angles by comparing the measurements with the expected gravity vector. However, accelerations generated by the tether will perturb these estimates and must be accounted for before using the measurements as part of the sensor fu-sion algorithm. Further, magnetometers are commonly used for yaw estimation by comparing the magnetic field with the expected direction of the magnetic field lines but suffers from inaccuracy (typically 1 − 5◦) and is prone to perturbations from local magnetic fields generated from the electronics and power cables. It is also possible to find the yaw angle using positional history from GPS mea-surements or combine two GPS sensors and compare the change in position for each sensor over time. However, most GPS signals tend to drop out when exceeding 4G’s (due to inherent design) and pose a major challenge for AWE systems that desire to operate at 10 − 15G’s or more. Kitemill has extensive experience with drifting in both po-sition and attitude estimates and proposes a novel solu-tion to pose estimasolu-tion that mitigates most of these chal-lenges.
Bluetooth 5.1 was announced 21 January 2019 and promises to provide centimeter positional accuracy [1] with a range of up to 1000 meters [2]. Using the angle of arrival and angle of departure enables the localization of Bluetooth devices and will result in a new array of prod-uct providing tracking of keys, cell phones, animals and other objects with high accuracy at a low cost (A Blue-tooth 5.1 chip costs about e4). Another potential lies in pose estimation for AWE systems, where this solution can reduce the levelized cost of energy through cheaper nav-igation systems. The basic idea is to place multiple Blue-tooth 5.1 devices inside the body of the kite oriented as its own reference frame representing the body frame of the kite. Then, position measurements can be used di-rectly to recreate the reference frame that describes the attitude of the kite.
This work models the Bluetooth 5.1 devices as particles that move with the kite, showing how to combine four parallel Unscented Kalman Filters and some simple equa-tions to find the pose of the kite. We share this work be-cause this solution has the potential to become a stan-dard for the AWE industry, however, it requires further R&D to create a complete system that is able to deliver the promised performance ś something that can be per-formed through a joint project between the companies.
References:
[1] Bluetooth. Enhancing Bluetooth Location Services with Direction Finding, 2019.
[2] https://blog.nordicsemi.com/getconnected/things-youshould-know-about-bluetooth-range
