Xiaomi Automobile Technology has published a patent for a method that identifies parking spaces even when painted lines are missing or obscured. The application, revealed in an IP database entry on March 11, describes a process that starts from “initial obstacle information,” groups those obstacles into a target set and then infers the boundaries of a parking space from that set.
The technique relies on environmental sensing and inference rather than explicit marking detection: by analysing nearby obstacles—other vehicles, curbs, poles or impediments—the system deduces where an available slot begins and ends. That approach complements vision-based line detection by filling a gap where lane markings are faded, covered by snow, dirt, leaves or other obstructions, and it can be implemented as part of a vehicle’s parking-assist or automated parking stack.
For consumers, the practical upside is straightforward: more reliable automated parking in real-world urban environments where parking lines are often degraded or informal spaces are common. For automakers and suppliers the patent is a small but telling sign that intelligent perception for low-speed manoeuvres is a competitive battleground, especially as companies race to ship user-facing features that make city driving less painful.
The filing also signals Xiaomi’s broader strategy to build an in-house intellectual property estate around vehicle software and sensing. Chinese OEMs and tech firms—ranging from traditional automakers to new entrants and internet companies—have been aggressively patenting autonomy-related techniques, creating a dense field of overlapping claims that will shape partnerships, supplier choices and potential litigation.
Technical limitations remain. Success depends on sensor quality, the robustness of obstacle classification and scene understanding, and validation in crowded, dynamic conditions. Nevertheless, a patent that formalises a method for inferring parking geometry without painted cues is a pragmatic piece of engineering that could be integrated quickly into consumer vehicles and parking-assist upgrades.