analiza aiqbench
This research replaces traditional hand-crafted models with a sparsely-gated Mixture of Experts network to solve bicycle localization in GNSS-denied environments. By integrating the MoE architecture directly into an Extended Kalman Filter, the system achieves superior accuracy while reducing computational cost by up to nine times compared to standard backbones. The approach demonstrates that data-driven representations can effectively generalize across diverse riders and road surfaces without sacrificing efficiency.
MoE-Based Learned Inertial Odometry for Bicycle Localization
arXiv:2510.17604v2 Announce Type: replace Abstract: 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.
Abstract: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 this http URL 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.
This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible
Robotics (cs.RO)
arXiv:2510.17604 [cs.RO]
(or arXiv:2510.17604v2 [cs.RO] for this version)
Text integral sub licența deschisă a publicației · 214 cuvinte
Subiecte conexe
MosaicIMU: Composing Carrier Experts for Generalizable Neural Inertial Odometry
arXiv · 9 iun. 2026
MI-SLAM: Magnetic Inertial SLAM Systems
arXiv · 27 aug. 2026
Contact-Aided Factor-Graph Localization for Underwater Sampling
arXiv · 28 aug. 2026
MARIO: Motion-Augmented Real-Time Multi-Sensor Inertial Odometry
arXiv · 3 iun. 2026