An exploratory clustering approach for extracting stride parameters from tracking collars on free ranging wild animals

Abstract

Changes in stride frequency and length with speed are key parameters in animal locomotion research. They are commonly measured in a laboratory on a treadmill or by filming trained captive animals. Here, we show that a clustering approach can be used to extract these variables from data collected by a tracking collar containing a GPS module and tri-axis accelerometers and gyroscopes. The method enables stride parameters to be measured during free-ranging locomotion in natural habitats. As it does not require labelled data, it is particularly suitable for use with difficult to observe animals. The method was tested on large data sets collected from collars on free-ranging lions and African wild dogs and validated using a domestic dog.

Biomechanics of predator–prey arms race in lion, zebra, cheetah and impala

“…To reduce noise, improve precision and increase temporal resolution in the position and velocity data, GPS and IMU measurements were fused 9 using a 12-state extended Kalman filter followed by a Rauch-Tung-Striebel smoother written in MATLAB (The Mathworks Inc., MA, USA). Data were segmented into strides, and non-gallop strides removed, as described in 42 and locomotor parameters determined for each stride. Stride data were separated into non-uniform speed bins with 400 data points in each and the 98th percentile of value determined for each bin.…”

“…Calculation of stride frequency Regression lines were fitted to stride frequency versus speed data at running speeds. Sections with running data were identified using an unsupervised clustering algorithm on three features derived from windows of accelerometer signals (4 s long) 42 . Features were chosen based on domain knowledge and were the standard deviation of the horizontal and vertical axis accelerometer signals and an autocorrelation estimate of the stride frequency 42 .…”

“…Sections with running data were identified using an unsupervised clustering algorithm on three features derived from windows of accelerometer signals (4 s long) 42 . Features were chosen based on domain knowledge and were the standard deviation of the horizontal and vertical axis accelerometer signals and an autocorrelation estimate of the stride frequency 42 . Features were normalised to have zero mean and unit standard deviation before they were clustered using the k-means algorithm.…”

Scaling of avian bipedal locomotion reveals independent effects of body mass and leg posture on gait

“…Yet, we hope we have highlighted the continuing utility of dynamic similarity and scaling principles for interpreting shifts in functional demands among animals of different size and morphology. Advances in technology have enabled measurement of locomotor dynamics over an increasingly broad range of conditions, including free-ranging and wild animals during foraging, predator-prey interactions and migration (Dewhirst et al, 2017;Hubel et al, 2016). Studies of non-steady locomotor dynamics can help reveal how animals balance multiple functional demands, including energetic costs, stability, injury avoidance, speed and maneuverability (e.g.…”

External validation of a collar-mounted triaxial accelerometer for second-by-second monitoring of eight behavioural states in dogs

“…Accelerometers, also called inertial sensors, have been used in horses for decades to detect and monitor lameness [ 11 – 14 ]. These sensors have also been used in horses for lying behaviour [ 15 ] and in wild animals for tracking behaviour [ 16 ]. In dogs, accelerometers have been used to monitor amount of activity [ 17 – 21 ], activity types [ 22 ], cognitive dysfunction [ 23 , 24 ] and for lameness detection [ 21 , 25 ].…”