This project uses the lighthouse technology or SteamVR tracking technology for developing low-cost position trackers. HTC uses the SteamVR tracking technology for their VIVE virtual reality system. Unlike video-based systems that use cameras, this system uses base stations or lighthouses which emit infra-red laser light. The laser light is intercepted by photodiodes which are mounted on the surface of an object called the tracker. The SteamVR tracking software can then estimate the bearing or the pose information of the trackers with respect to the base stations by recording the timestamp information of when the photodiodes get hit with the laser light. The timestamp information along with the relative photodiode distance information is then used to estimate the spatial position and orientation of the position trackers. Along with optical tracking this technology also uses IMUs for dead reckoning. The system is quite analogous to the GPS navigation systems where multiple satellites help triangulate the position of a GPS receiver based on highly accurate timestamp information.

The current works with the technology focus on developing compact and body mountable position trackers for recording human biomechanics. Having position trackers strapped on to a persons limb can lead to skin and tissue motion artifacts affecrting the tracking accuracy. Key to the optimization routine was reducing the mass and the inertial effexts of the tracker without affecting tracking performance. The tracking performance is dictated by the location of the photodiodes on the tracker. This work entails solving the design optimization problem where we intend to find optimal positions and orientations for the photodiodes to have the best tracking performance results.

The goal was to optimize the geometry of the tracker defined by the placement of photodiodes. To optimize the geometry or photodiode locations, it is important to parameterize the locations and formulate an objective function for tracking effectiveness. An evaluation criterion was formulated to determine the effectiveness of tracking. Two types of optimization-based works were explored:

1. Monte Carlo Simulations

2. Gradient Descent based Constrained Optimization

Since the tracking performance is dictated by the position and orientation of photodiodes on the tracking surface these were the variables of optimization. A cost function was formulated to estimate the number of hits based on the position and direction of photodiodes. Since, the photodioes have finite field of view the evaluation function can estimate which photodiode is getting hit based on the line of sight vector from the base station to the photodiode. The optimizer traies to maximize the number of hits or the tracking score by adjusting the photodiode position and orientation. Post optimization a tracker was designed based on the optimum solution in a cuff-like geometry that reduced mass by 20 gm and improved tracking performance by 8-10 % compared to the VIVE tracker.