Abstract
Camera-based motion tracking systems, which fuse camera images and inertial measurement unit data to estimate a position and orientation of a device, can experience performance bottlenecks on resource-constrained platforms due to processing a statically defined number of visual constraints. A dynamic, map-aware framework can allocate a visual constraint budget at runtime. When a map is available, the framework may evaluate a quality of the map and choose to use map-based tracking features instead of standard motion-tracking measurements. A budget for persistent feature tracking landmarks may optionally remain fixed. This approach can reduce a total number of processed constraints. Modifying a constraint budget dynamically may conserve computing resources, lower power consumption, and improve processing latency for the device.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Agrawal, Parth and DuToit, Ryan Christopher, "Changing Computing Resource Limits for Camera-Based Motion Tracking", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/12393