By Remco Sikkema
In my 15 years with Xsens, I have witnessed the evolution of one of the most advanced motion capture systems ever built. Through this journey, we have faced and overcome numerous technical challenges. While no technology is without flaw, we strive to make our motion capture system as close to perfect as possible.
Our mocap systems feature sensors worn directly on the body – either embedded in a suit or attached using straps. Placed strategically on key points of the human body, these sensors measure movement without relying on external cameras. This offers unmatched freedom and flexibility to users, allowing them to simply put on the Xsens suit, and capture motion in any environment. However, there is one drawback to inertial motion capture known as ‘drift’, and for us to deliver our promise of effortless mocap, we had to take care of it.
What is drift?
Drift occurs when, after recording a movement that ends at the same location as it began, the motion capture software indicates a different starting point. Essentially, the virtual representation of the movement doesn’t align perfectly with the movement in the real world. Research has shown that drift is more prominent during fast, dynamic movements such as running, and when the user is not touching the ground, such as during jumps or fight scenes. On the other hand, slow and deliberate movements result in less drift.
Why does it happen?
Due to a lack of external references such as cameras, inertial mocap lack absolute positioning. They rely on mathematical calculations to determine position. These calculations tend to introduce minor errors and as the errors accumulate over time, there is noticeable drift in positioning. Drift is inherent to inertial mocap and eliminating it completely is a challenge. However, at Xsens, we have made remarkable strides in bringing it down to a minimum.
How did we solve it?
Our advanced sensor fusion algorithms are key to our success in minimizing drift. With decades of experience and expertise embedded in our technology, we have leveraged machine learning (ML) and artificial intelligence (AI) to optimize our algorithms for the best possible result. We continuously refine these algorithms, especially for scenarios most prone to drift, such as fast movements or activities with low ground contact. This is significantly enhanced the precision and reliability of our system.
In this video below, you'll see Xsens motion data from a person starting at a designated point (marked by a red arrow), engaging in various activities like walking, running, and even using stairs. At the end of the video, the person returns to the starting point, and remarkably, there's virtually no drift.
Our achievement in minimizing drift is unmatched in the industry. No other inertial mocap product on the market can demonstrate such a result.
Absolute Positioning: HTC Vive Integration
While our current system is great at minimizing drift, some situations, like virtual production, need absolutely precise positioning. That's where Xsens Animate's native integration with HTC Vive comes into play. By combining motion data from Xsens Link or Awinda with HTC Vive's position tracking, we can significantly improve absolute position accuracy.
This means you can track people, objects, and props with pinpoint accuracy within the MVN Animate engine. Simply attach Vive Trackers to anything you want to track—people, tools, boxes, sports gear—and you'll get precise position and motion data no matter where you are. You can read all about this integration here.
Xsens mocap
Xsens helps animators to create realistic movements for characters and by choosing Xsens, you’re not just investing in technology, but also in the future of your studio, your team, and your storytelling.
Explore the world of Xsens motion capture by visiting the MVN Animate page, or you can request more information via this link.