Table of Contents

Multi-period fusion solution

Multi-period fusion positioning is an advanced feature designed to address the persistence of Mega experiences in complex lighting environments. By constructing an atlas covering different time periods, the system can overcome interference from day-night alternation and seasonal changes on visual features, ensuring centimeter-level precise positioning at any time of day.

Core challenges

Mega primarily relies on environmental visual features for positioning. Although the algorithm has been optimized for lighting and seasonal variations, drastic illumination differences caused by day-night alternation may fundamentally alter the visual features of the same location. Therefore, map data collected at a single time point (e.g., only daytime) often fails to match during another period (e.g., nighttime) due to significant feature differences, resulting in positioning failure.

Solution

To solve all-weather positioning challenges, the Mega platform provides multi-period data fusion capabilities. By processing data from different periods in the cloud, the system's adaptability to lighting changes is further enhanced.

How it works

  1. Multi-period collection: Collect data for the same physical scene under representative different lighting conditions (e.g., daytime, nighttime).
  2. Cloud data fusion: Upload all collected data to the Mega cloud, where cloud services automatically process multi-period data. Through feature fusion optimization, a map database containing different periods is constructed.
  3. Automatic matching and positioning: During application runtime, the system automatically retrieves and matches the map closest to the current lighting conditions from the fused map data based on real-time camera-captured features, returning the image's pose in the map.

Best practices

To achieve optimal fusion results, follow these collection specifications:

  • Cover key periods: Include at least one set of "daytime" and one set of "nighttime" data. For scenes with extreme lighting changes (e.g., squares with landscape lights on a fixed schedule), add "pre-lighting" and "post-lighting" collections.
  • Path consistency: Although collected at different times, maintain consistent walking paths and shooting angles during each collection. This helps the cloud perform feature alignment and fusion more efficiently.

Implementation process

To enable multi-period fusion positioning, follow a specific collection and configuration workflow.

  1. Collection planning

    Evaluate lighting variations in the scene to determine required period combinations:

    • Basic combination: One daytime dataset + one nighttime dataset (recommended after full streetlight activation)
    • Enhanced combination: Add dusk data if special lighting exists during high-traffic twilight hours

  2. Data collection

    Ensure consistent walking paths and shooting angles across periods (e.g., if daytime collection follows a south-north centerline path, maintain identical route at night). This helps the cloud efficiently compute geometric relationships between period-specific maps and significantly improves alignment accuracy.

    Before collecting multi-period data:

  3. Map construction

  4. Review mapping results

    After mapping completion, review results including collection routes and spatial models:

    Tip

  5. Test positioning performance

Important

Reminder: Maintain consistent walking paths and shooting angles during each period's data collection. This helps the cloud efficiently compute spatial relationships between sub-regions and improves map alignment accuracy.

Tip

Multi-period maps are optimized fusion results with strict alignment. Annotations only need placement once.