用稀疏门控专家网络提升自行车定位精度与效率
MoE-Based Learned Inertial Odometry for Bicycle Localization
- 采用稀疏门控专家网络构建数据驱动的惯性里程计，替代传统手工模型
- 在多骑行者、多路面场景下实现0.333米/秒速度误差，9.49米ATE，2.58米RTE
- 仅28.76M FLOPs，计算量仅为同类模型的1/7至1/9，适合边缘部署
GNSS在城市峡谷中受多径误差影响，难以实现可靠的自行车定位。传统手工设计的惯性方法（如骑行航位推算、非完整约束）无法泛化到不同骑行者、姿态和路面。本文提出一种基于学习的惯性里程计：利用多骑行者、多自行车、多路面的IMU数据训练神经速度预测器，与扩展卡尔曼滤波器紧耦合，实现三维位姿估计，以数据驱动方式替代手工模型。为降低计算开销，预测器采用稀疏门控混合专家（MoE）结构，结合Top-K路由与参数冻结训练策略，并设置每专家容量限制以促进多样化分工。所提模型在推理时误差为0.333米/秒，平均轨迹误差（ATE）为9.49米，路径误差（RTE）为2.58米，仅需28.76M FLOPs。相较LLIO与TLIO所用的ResMLP和ResNet骨干网络，计算量减少约7倍和9倍，同时保持或超越其精度。在所有评估划分上，该模型的MoE变体在ATE与RTE指标上均排名第一，表现出强泛化能力。
原文摘要 · Abstract (English)
GNSS suffers from multipath errors in urban canyons, making reliable bicycle localization difficult. Hand-crafted inertial alternatives, such as cycling dead reckoning and nonholonomic constraints, fail to generalize across riders, postures, and road surfaces. This work implements learned inertial odometry for bicycle localization: a neural velocity predictor, trained on multirider, multi-bicycle, and multi-surface IMU data, is tightly coupled with an extended Kalman filter for three-dimensional pose estimation, replacing hand-crafted cycling models with a data-driven representation. To reduce computational overhead, the predictor is realized as a sparsely-gated Mixture of Experts (MoE) network with top-K routing, trained under an alternating parameter-freezing scheme with a per-expert capacity constraint to encourage diverse and balanced specialization.The proposed model attains 0.333 m/s inference error, 9.49 m ATE, and 2.58 m RTE at only 28.76 M FLOPs. This corresponds to approximately 7 times and 9 times fewer FLOPs than the ResMLP and ResNet backbones used in LLIO and TLIO, while maintaining comparable or superior accuracy. The proposed MoE variants further rank first on both ATE and RTE across all evaluation splits, demonstrating strong generalization.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。