Ultra-lightweight 6-directional odometry cuts drone position error by half
TIO-Former: Ultra-Lightweight 6-Directional ToF-Inertial Odometry for Nano-UAVs via a Streaming Causal Transformer
Robotics
Summary
Small drones need to know where they are to fly safely but carry very limited sensors and computers. The authors developed TIO-FORMER, a way to combine signals from tiny time-of-flight sensors and motion sensors using a special transformer model that runs fast on small chips. This method does not rely on cameras or maps and works well even with limited data and tricky environments. In tests, it cut position errors by over half compared to existing methods and runs quickly on low-power drone electronics.
What this means in practice
- •For drone developers: Implement accurate drone position tracking using lightweight sensors and transformer models suitable for tiny drones with tight resource limits.$Commercial implications: Enables commercial production of autonomous nano-UAVs that navigate precisely on minimal hardware, improving delivery, inspection, or surveillance robots.
- •For robot integrators: Integrate efficient range and inertial odometry for robots operating where cameras or LiDARs are impractical due to payload or lighting constraints.
Authors
Yang Liu, Yifan He, Wenhao Zhao, Xiangyu Mo, Yang Xu, Hao Wei, Mingze Ma, Huan Li, Yifan Wu, Zipeng Dai, Xin Zhou, Fei Gao
Abstract
Autonomous nano-UAV navigation requires accurate ego-motion estimation under stringent size, weight, power, and computing (SWaP-C) constraints, where visual sensors and LiDARs exceed payload limits, optical flow degrades in low-texture scenes, and inertial-only state estimation is susceptible to accumulated drift. While multi-zone time-of-flight (ToF) arrays provide a lightweight metric complement, 6-DoF estimation from merely 384 ranges per frame is challenged by invalid returns, anisotropic observability, and temporal computational scaling. We propose TIO-FORMER, a camera-free, optical-flow-free, and mapless range-inertial odometry framework driven by an IMU and an ultra-lightweight (15 g) payload of six orthogonal 8 x 8 ToF arrays. Our frontend pairs consecutive range grids with a bilateral gated difference, while IMU-guided cross-attention dynamically routes directional features conditioned on platform kinematics. A Streaming Causal Transformer couples an uncompressed Local KV cache with compressed Chunk-FIFO memory, maintaining bounded inference cost and memory footprint independent of flight duration. In real-flight evaluations, TIO-FORMER reduces open-loop position error by 54.4% compared to nano-UAV optical flow and by 66.4%-89.1% over learned inertial baselines. We also evaluate performance across multiple environments and robustness under severe sensing degradation. Deployed on an edge RISC-V companion computer, TIO-FORMER achieves a P95 latency of 10.466 ms and peak resident memory of 6.324 MiB (less than 5 percent system RAM), demonstrating that sparse range sensing provides practical geometric anchoring for resource-constrained micro-aerial robots. Code is available at https://github.com/Ly041021/TIO-Former.