Ringnav improves heading estimates for tiny robots with circular memory
RingNav: Task-Shaped Circular Memory for Visual Heading
RoboticsComputer Vision and Pattern Recognition
Summary
Tiny robots have very limited memory and computing power, which makes navigation hard. The authors developed RingNav, a small and efficient system that remembers how much a robot turns using a special circular memory. This helps the robot figure out its direction more accurately and works well even when visuals are unclear. RingNav runs fast on tiny processors suitable for very small robots like drones.
What this means in practice
- •For robotics engineers: Build lightweight visual heading estimators that reduce turning errors for small mobile robots with limited computing power.
- •For drone developers: Deploy efficient heading estimation algorithms on ultra-low power processors for drones operating in visually degraded environments.
Authors
Naitri Rajyaguru, Aryan Tajne, Stephen Jardim, Cornelia Fermuller, Yiannis Aloimonos
Abstract
Tiny robots have limited computation and memory, yet visual navigation systems often maintain more state than a task requires. We ask whether persistent memory can instead be shaped by the task. We present \textbf{RingNav}, a \(10.6\)k--parameter visual heading estimator that constrains its memory to the circular manifold \(S^1\). RingNav learns only the turn between consecutive images, while heading is composed directly on the circle. On the TartanAir Visual SLAM Challenge, RingNav reduces mean absolute turn error by \(27\%\) and accumulated heading error over $4$ sec by \(15\%\) relative to a similarly sized learned-memory model. The structured update also gives exactly zero turn for identical frames and exact cancellation under sequence reversal. Without fine-tuning, RingNav transfers to real-world drone experiments and continues estimating heading under severe visual degradation. RingNav runs at $27.3$Hz on Raspberry Pi Zero 2W and can be deployed on an ultra-low power GAP8 processor at $33.4$Hz.