PhD Thesis Proposal: Karan Ahuja, "Practical and High-Fidelity User Digitization On-the-Go"

 

As we build towards our vision of smarter environments, we are faced with the challenge of intuitive, rapid and natural interaction across a multitude of computational ecosystems. This requires us to not only seamlessly interact with our environment but accurately digitize ourselves within it. This remains challenging due to the multifarious design space of sensing modalities and their placements. In general, sensors placed on a particular joint only estimate the pose of that particular limb. For example, commercial AR/VR systems such as the Oculus Quest use the headset and handheld controllers to track those particular limbs. Alternatively, motion capture technologies -- such as Vicon optical tracking and XSens IMU suits -- increase the fidelity of body capture, but decrease the practicality in lockstep, to the point where it seems unlikely for body digitization to become ubiquitous.

In contrast, my research aims at exploring pathways to high fidelity full-body digitization with minimal user instrumentation, or ideally, no instrumentation at all. I achieve this by adding layers of intelligence over existing practical and ubiquitous sensing platforms, thus enabling them to be the drivers of high fidelity user digitization. Complimentarily, I also focus on making high-fidelity systems more practical by decreasing their points of instrumentation, potential invasiveness and privacy implications. Finally, for my PhD, I propose two projects that extend my vision and make further inroads towards the idealized vision of practical high-fidelity user digitization.