Tests confirm an impressively low 6-7 milliseconds real-time latency. At a 200 Hz data capture this equates to just a fraction over a one frame delay in real-time data capture to process and display of data.
ETH measured movements simultaneously using an incremental wire potentiometer (16 cts/mm, sampling at 1000 Hz), and a ten-camera Oqus-300 system running at 200 Hz. The wire potentiometer was directly linked to an xPC target; latency was assumed to be negligible. Marker positions were also received at the xPC target via UDP from a computer running Qualisys Track Manager, 2.1.
Both data streams were sent from the xPC target to a control host where latency was quantified. The monitored movement was based on the sliding of a seat in a rowing boat, repetitively performed forward and backward over a distance of 0.6 m at about 1.6 Hz. Movement was performed along one axis of the global coordinate system defined for motion capture, thus, the related coordinate of the tracked object representing the seat’s movement (with markers placed on it) was compared to the potentiometer also measuring the seat’s movement. The experiments were repeated with additional, static objects which were also tracked.
Latency of the motion capture system was found at our control host, i.e. at the end of the chain. Depending on the number of static objects, latency varied from six to seven milliseconds.