Reproduce “TCB-VIO: Tightly-Coupled Focal-Plane Binary-Enhanced Visual Inertial Odometry” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Matthew Lisondra, Junseo Kim, Glenn Takashi Shimoda, Kourosh Zareinia, Sajad Saeedi

TCB-VIO is presented as the first tightly-coupled Visual-Inertial Odometry (VIO) system designed for Focal-Plane Sensor-Processor Arrays (FPSPs), demonstrating superior accuracy and robustness in dynamic environments. The system achieves significantly reduced computational latency and energy consumption by performing on-sensor binary feature extraction, with edge detection showing a 16.4x speedup over CPU and consuming thousands of times less energy.

Key insights

  • Integrating sensing and processing directly on FPSPs dramatically reduces latency and power consumption, enabling high-frame-rate VIO suitable for real-time applications in agile robotics.
  • Implementing a tightly-coupled VIO architecture on FPSP hardware provides superior accuracy and robustness compared to loosely-coupled alternatives, especially in challenging, dynamic scenarios.
  • The use of binary-enhanced features with a feathered edge map and a specialized KLT tracker effectively leverages FPSP capabilities, maintaining visual tracking robustness while offloading computational burden from the host processor.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.