Perfection in Motion
In competitive sports, nuances in the course of movement can decide on victory or defeat. For this reason, professionals, trainers and sports physicians fine-tune the interplay of arms, legs, torso and head with scientific meticulousness. Not only to improve physical performance, but also to prevent injuries. A common method for analysing movements is motion capture. In most cases, reflective markers are used, which are attached to test persons. The Munich-based company Simi Reality Motion Systems uses Artificial Intelligence to enable markerless 3D motion capture.
A sound barrier that was long considered to be set has been broken: A man has run a marathon in less than two hours. However, this was less an outstanding individual achievement than a project planned down to the last detail: the run took place under precisely defined conditions that have little to do with a regular race. Nothing was left to chance, from the course layout to the nutrition and the equipment. And of course, the running style played a role, if not the decisive one. In order for an athlete to achieve such a performance, the optimum of his or her body characteristics must be achieved with every movement.
For many years now, such sports science analyses have been carried out on the computer. What has changed is the way in which the movement information reaches the computer. Up to now, reflective markers were necessary, which were attached directly to the person to be examined - this method is called Motion Capture. The American film industry uses this method to bring fictional creatures such as King Kong or Gollum (Lord of the Rings) to life. The output data is generated by actors and actresses whose movements are transferred to a computer system using the aforementioned markers and infrared cameras. What works convincingly in Hollywood studios, however, could not be used in real environments such as in the field of sports until now.
Markerless tracking based on Deep Learning
This is where the motion detection system of the Munich-based company Simi Reality Motion Systems GmbH comes in. It does not require markers. Instead, the analysis is based on images which can also be generated outdoors, for example. The movements are recorded and stored using synchronized industrial cameras which offer a very high frequency and high resolution. Using the latest algorithms from image processing, the Simi Motion program is able to record the movement in detail and display it in the form of 2D or 3D data. This markerless tracking method is based on approaches from deep learning (including convolutional pose machines).
In simple terms, the AI application helps the computer to understand the three-dimensional movements of humans by means of camera images - in the same way as the human brain does with its eyes. The concept was developed together with the Leibnitz University of Hanover. The application has been on the market since 2016 and is used at world championships and in professional leagues.
Markerless Motion Capture is also gaining importance in other areas - for example, augmented reality or virtual reality in autonomous mobility or medicine. In addition, application possibilities are opening up in industry (e.g. in process optimization) as well as in services, for example when data is processed within the framework of cloud service contracts.
Robotics and Autonomous Systems
- Intelligent Sensor TechnologyIntelligent Automation
Simi Reality Motion Systems