Optical tracking in hypersonic wind tunnels

We’ve developed the capability to use two independent cameras (at different frame rates) to extract aerodynamics from free-flight models in our TUSQ hypersonic wind tunnel.

All image processing and analysis is performed in open-source code, making use of open-cv image processing functions. The primary challenge with this project is reducing the measurement noise which gets amplified when the position and orientation of the body is differentiated twice to extract forces and moments. To resolve this, we use a custom-written Unscented (Sigma point) Kalman filter and Bayesian smoother for state estimation, which provide smooth second kinematic derivatives.

See some of our videos below, and check out our papers for further information!