Unmet Need
Marker-based 3D optical tracking technology is used in multiple markets, the two largest ones being 3D Motion Capture and Surgical Navigation. The 3D motion capture market has an estimated size of $151.8 million (Verified Market Research).
In state-of-the-art systems, motion tracking is either done optically or electromagnetically (EM). EM trackers are inaccurate in environments containing ferromagnetic materials or when the markers are far from the sensor, and require connected (wired) markers. Optical trackers only require that the passive markers are in direct line of sight of the camera(s), which makes them the preferred tracking method in most applications.
Currently, there are two predominant options for conducting marker-based optical tracking: multi-camera systems with point like markers and single camera systems with structured markers.
Multi-camera systems are capable of tracking the position of small spherical markers from a wide range of fields-of-view at high precision but require an expensive array of sensors, specialized hardware and complicated setup and calibration. Orientation information can be inferred from the 3D positions of multiple tracked markers. Such systems are used in most human motion capture applications.
Existing single camera systems track larger markers that feature sub-structures (e.g. QR tags) to enable the calculation of 3D orientation from a single camera view. These provide precise measurement only at limited range of orientations and positions; therefore they are not suitable for most applications where the targets can move around in freedom in the environment. Such single camera systems are often used in industrial applications where the line-of-sight and view angles are predictable or well controlled.
Surgical tracking system typically use multiple cameras and structured markers in order to prevent line-of-sight occlusions and to extend the range of detectable orientations.
Animal tracking is a special application space where the targets (animals) tend to move around in a way that necessitates the ability to track a wide range of orientations. In current practice it means that most 3D animal trackers use expensive and complicated multi-camera setups.
Currently there are only two sections of the market that are well served by existing technologies: expensive and complicated multi-camera systems, and cheap systems with limited accuracy and functionality. There is a strong need for a solution that works on cheap hardware that is simple to set up, yet capable of providing precise measurements over a wide range of orientations and positions.
Technology Overview
Researchers at Johns Hopkins have developed a system combining hardware that can be commonly found in behavioral laboratories (a single camera, a ring light mounted on the camera, and a computer) with a specially-designed 3D marker to be placed on the animal and a computer vision algorithm that can process the camera images and generate the animal’s 3D position and orientation in each frame. The system has an accuracy comparable to a popular commercial system and a wider range of angles than a popular open-source system.
In addition to the commercial-level accuracy accomplished at very little cost, this invention’s generalizability is also compelling. The system can be adapted for use on bigger animals (which would require a larger marker and different camera) and objects such as drones. The invention could even be applied to multiple targets simultaneously, which could enable analysis of interactions.
Stage of Development
A prototype of this invention has been built and used in laboratories.
Publication
Vagvolgyi, B. P., Jayakumar, R. P., Madhav, M. S., Knierim, J. J., & Cowan, N. J. (2022). Wide-angle, monocular head tracking using passive markers. Journal of Neuroscience Methods, 368, 109453.
