Minimalist Visual Inertial Odometry

Francesco Pasti, Jeremy Klotz, Nicola Bellotto, Shree K. Nayar

arXiv:2605.19990 · 2026-05-18 공개 · arXiv · PDF

robot-navigation visual-inertial-odometry minimalist-sensing photodiode gabor-mask temporal-convolutional-network planar-odometry differential-drive

Abstract

Visual-Inertial Odometry(VIO), which is critical to mobile robot navigation, uses cameras with a large number of pixels. Capturing and processing camera images requires significant resources. This work presents a minimalist approach to planar odometry, demonstrating that just four visual measurements and an IMU can provide robust motion estimation for differential-drive robots. Our key insight is that four downward-facing photodiodes that sense the world through optical Gabor masks produce signals that encode speed. Based on this, we jointly optimize the mask parameters alongside a Temporal Convolutional Network (TCN) using a physically-grounded simulator. The resulting model decodes speed from just the four measurements produced by the photodiodes. Pairing these estimates with the angular speed from an IMU yields a continuous planar trajectory. We validate our approach with a prototype sensor mounted on a differential drive robot. Across diverse indoor and outdoor terrains, our system closely tracks the reference ground truth without any real-world fine-tuning. Our work shows that minimalist sensing enables efficient and accurate planar odometry.

한국어 요약

📋 한 줄 요약

**[Visual-Inertial Odometry / Minimalist Sensing]** 4개 photodiode + Gabor mask + IMU만으로 differential-drive 로봇의 planar odometry — TCN과 mask 파라미터를 simulator로 joint 학습, 실내·실외 다양 지형에서 real-world fine-tuning 없이 ground truth 추적.

🎯 핵심 기여도

💡 핵심 아이디어

모바일 로봇 planar odometry는 픽셀 군이 필요하다는 통념과 달리 optical Gabor mask를 통해 sense하는 4개 photodiode 신호로도 speed를 encode할 수 있으며, mask 디자인과 TCN을 simulator로 함께 학습하면 minimalist sensing으로 robust planar motion estimation이 가능하다.

🔬 기술적 접근법

📊 주요 결과

💭 의의 및 한계

**의의**: 로봇 sensing의 minimalist 패러다임 — resource·cost·전력 절감, simulator에서 mask·network 공동 최적화의 design 패턴, real-world fine-tuning 없이 sim-to-real. **한계**: Planar(2D) 동작에 한정·3D motion 미커버, differential-drive 로봇·downward-facing 가정, 매우 다양 광학·노이즈 환경에서의 robustness는 추가 검증.

🚀 실용적 활용