Huawei’s Xingji (星迹) is a scenario-specific positioning system that combines satellite measurements, specialized hardware, AI motion estimation and environmental/map matching, rather than relying on GNSS alone. Reports of Li Xiaolong’s October 12, 2026 briefing describe its initial implementation on the Mate 90 series and WATCH GT Runner 2, targeting situations where conventional positioning becomes inaccurate or unavailable. 68

• Elevated roads — more precise ranging and lane identification. Huawei combines HiSilicon short-range chips with GNSS carrier-phase measurement, claiming more than tenfold better ranging accuracy and sub-meter positioning for lane-level navigation; the intended benefit is sufficiently precise guidance to change lanes on complex elevated roads. The tenfold figure concerns ranging, not a demonstrated tenfold improvement in every navigation scenario. 57

• Tunnels — an alternative to inertial-only tracking. AI speed estimation plus geomagnetic matching is intended to maintain positioning when satellite signals are blocked, addressing the accumulating error of conventional inertial navigation. Huawei describes this as industry-exclusive tunnel lane-level positioning that is unaffected by tunnel length or driving time; those are Huawei’s reported claims, not independently established guarantees. 57

• Running tracks — lane-specific guidance on the phone, improved reception on the watch. The Mate 90 series combines RTK—real-time kinematic positioning—with AI-PDR, AI-based pedestrian dead reckoning, to support running-track lane-level navigation. Separately, the WATCH GT Runner 2 uses a 3D floating antenna, with Huawei claiming GNSS signal strength 5.4 dB higher than on the first-generation running watch, addressing inaccurate tracks in complex running environments. 69

• Ski slopes — motion-specific trajectory reconstruction. GNSS/SkiDR fusion combines satellite positioning with ski-specific dead reckoning to address distorted tracks during high-speed skiing; Huawei reports fivefold finer trajectories, twice the dynamic responsiveness and 98% ski-distance accuracy. 9 Insufficient evidence was returned to verify the additional claims of maps covering more than 3,000 resorts and over 97% ski-run matching accuracy.

• Underground parking — record both floor and parking position. 3D road-network matching is used to record the parking location, with reported floor-identification accuracy above 95% and position error under 10 meters. Mate 90’s Precise Car Finder 2.0 is reported to improve parking-position accuracy by 50% versus Mate 80 and cover more than 30,000 underground garages. 75

• Shopping malls — indoor location services plus mapped destinations. Huawei and partners extend location-based services into commercial interiors, reporting coverage of 500,000 commercial-district geofences and 17,000 indoor-positioning buildings. Petal Maps separately offers store-level navigation in more than 260 malls, with more than 200 additional malls expected each month—the latter is an expansion forecast, not existing coverage. 5

The important qualification is that these are different scenario-specific capabilities, not one universal sub-meter positioning promise. I don’t know the complete test conditions or comparative baselines, and there is insufficient evidence here to independently validate the performance figures or tunnel-exclusivity claim.

Huawei’s Xingji (星迹) is a scenario-specific positioning system that combines satellite measurements, specialized hardware, AI motion estimation and environmental/map matching, rather than relying on GNSS alone. Reports of Li Xiaolong’s October 12, 2026 briefing describe its initial implementation on the Mate 90 series and WATCH GT Runner 2, targeting situations where conventional positioning becomes inaccurate or unavailable. 68

• Elevated roads — more precise ranging and lane identification. Huawei combines HiSilicon short-range chips with GNSS carrier-phase measurement, claiming more than tenfold better ranging accuracy and sub-meter positioning for lane-level navigation; the intended benefit is sufficiently precise guidance to change lanes on complex elevated roads. The tenfold figure concerns ranging, not a demonstrated tenfold improvement in every navigation scenario. 57

• Tunnels — an alternative to inertial-only tracking. AI speed estimation plus geomagnetic matching is intended to maintain positioning when satellite signals are blocked, addressing the accumulating error of conventional inertial navigation. Huawei describes this as industry-exclusive tunnel lane-level positioning that is unaffected by tunnel length or driving time; those are Huawei’s reported claims, not independently established guarantees. 57

• Running tracks — lane-specific guidance on the phone, improved reception on the watch. The Mate 90 series combines RTK—real-time kinematic positioning—with AI-PDR, AI-based pedestrian dead reckoning, to support running-track lane-level navigation. Separately, the WATCH GT Runner 2 uses a 3D floating antenna, with Huawei claiming GNSS signal strength 5.4 dB higher than on the first-generation running watch, addressing inaccurate tracks in complex running environments. 69

• Ski slopes — motion-specific trajectory reconstruction. GNSS/SkiDR fusion combines satellite positioning with ski-specific dead reckoning to address distorted tracks during high-speed skiing; Huawei reports fivefold finer trajectories, twice the dynamic responsiveness and 98% ski-distance accuracy. 9 Insufficient evidence was returned to verify the additional claims of maps covering more than 3,000 resorts and over 97% ski-run matching accuracy.

• Underground parking — record both floor and parking position. 3D road-network matching is used to record the parking location, with reported floor-identification accuracy above 95% and position error under 10 meters. Mate 90’s Precise Car Finder 2.0 is reported to improve parking-position accuracy by 50% versus Mate 80 and cover more than 30,000 underground garages. 75

• Shopping malls — indoor location services plus mapped destinations. Huawei and partners extend location-based services into commercial interiors, reporting coverage of 500,000 commercial-district geofences and 17,000 indoor-positioning buildings. Petal Maps separately offers store-level navigation in more than 260 malls, with more than 200 additional malls expected each month—the latter is an expansion forecast, not existing coverage. 5

The important qualification is that these are different scenario-specific capabilities, not one universal sub-meter positioning promise. I don’t know the complete test conditions or comparative baselines, and there is insufficient evidence here to independently validate the performance figures or tunnel-exclusivity claim.