|Johansen, Tor Arne
|Vedeler, Alexander Georg
|Motivated by the lack of support for snowkiting in modern sports trackers, this thesis
investigates height estimation of snowkiting jumps. Since a jump with a kite does not
involve free-fall due to the pull of the kite, traditional jump algorithms cannot be used.
By utilizing a consumer grade MEMS barometer-IMU, a sensor fusion is implemented in
order to estimate jump trajectories and present the user with performance values such as:
jump height, drop and airtime.
Due to the harsh and windy conditions of snowkiting barometric sensor data can be noisy
and therefore a feed forward indirect Kalman filter with a RTS smoother is implemented
and compared against a direct Kalman filter with a RTS smoother. By mounting the sensor
directly on the snowboard or ski an algorithm was developed in order to detect the exact
of takeoff and landing.
Testing in the field showed that barometers are highly sensitive to wind and waterproof
trackers needs more than one ventilation hole, otherwise an increasing pressure can build
up inside the sensor. With an unreliable barometric sensor the feed-forward indirect filter
showed good performance on single jump data sets without drifting or being too affected
by noisy barometric data. Due to its complementary design it utilizes the accelerometer
during sharp transients such as takeoff and landing while trusting the barometer before
takeoff and after landing for estimating initial and final altitude of the jump. The proposed
solution achieved a jump height RMSE of 0.48m over 33 jumps, with jumps varying from
1.56m to 7.15m.
|Kybernetikk og robotikk, Autonome systemer
|Tracking height of snowkiting jumps based on MEMS barometer-IMU sensors
|Norges teknisk-naturvitenskapelige universitet, Fakultet for informasjonsteknologi og elektroteknikk,Institutt for teknisk kybernetikk