2020 Work Log

Today, the main tasks were as follows:

Completed synchronizing the MDM proto and MDM spec

1. Worked with Lin Nan to correct some wording errors in the MDM spec if you noticed them.
2. Left several remaining items that Lin Nan and I couldn’t decide on alone.
3. Decided during the 2020-04-26 morning meeting to postpone the most troublesome LaneSplitMerge and Polyline changes.
4. Completed modifications to the proto.
5. Completed modifications to nexus.
6. Completed modifications to mdk.
7. Also fixed a test data issue (crazy data error reporting).

Bug fixes

With the April 30 release approaching, the number of bugs surged.

HDMAPMDK-1111 analysis completed

• Automatic test case copy script
• Export to Apple Reminder

• Completed object lane POI binding and finished debugging [4]

Opposite Lane POI Binding

• Perform binding on data after multidigitization
• Construct test data

RoadBorder Fix

• Completed migration of test cases
• Improved Dune and its ReadMe to directly output test data Had a chat with Weiyu, looking ahead, I will focus on the compiler system.

Opposite Lane POI Binding

• Perform binding on the data after multidigitization

Custom Rate Limiting

• Guangcong turned the input into a vector and added tests
• One minor issue: Guangcong doesn’t fully understand how C++ pointers work
• Maybe try to fix https://github.com/scalameta/scalafmt/issues/337?
• Share the files Uncle provided as anonymous links via Resilio Sync

RoadBorder Fix

Discovered that I had made wrong assumptions. The lc and rb information recorded in rc may have the following cases:

1. There is only one lc, so we cannot assume that there are at least two lc entries.
2. The rb on one side might not exist, so we cannot assume that rb always exists.

Completed rate limiting.

• SpeedLimit compilation verification: OSMSerializable is not good because the bound SpeedLimit has no geometry There may be problems: speed limit binding during ramp merging may cause issues
• Interpolation
• Hausdorff distance
• LcInJunctionTile

Some Ingenious Ideas

• Use Spark directly?
• What does OSM’s Context need?

• Clarified with Yiming about the custom speed limit interface: he guarantees that the input location pair can always determine a unique shortest route. After returning, sorted out with Guangcong the algorithm for generating custom speed limits:
  1. Given a route
  2. Given that |start DiRoadOnPath::count - end DiRoadOnPath::count| ≤ 1, then we can always find a pair (start DiRoadOnPath, end DiRoadOnPath) such that the distance is minimized.
• Temporarily released a usable snapshot version of the pylon-v0.1.1 compiler for Xiaokang.
• Went downstairs with Hanteng to wander around and shoot the breeze.
• Clarified with Zizhe and Yicheng about the SUMO map usage interface.
• The speed limit compilation succeeded but has not been verified yet.

• Determined custom rate-limiting scheme @guangcong
  1. Used the new module-dynamicEvent to solve this problem
  2. Custom rate-limiting updates are published as a new event to trigger horizon data updates
  3. Need to consider exception handling and memory issues
     After discussing with Guangcong, he encountered some issues
• Xiaokang’s review
  Java, as a solid language, still has a lot to offer
• Rate-limited compilation
  Basically no progress, added logic for extracting RoadCenterTile
• Sectioning issue resolved, analyzed edgeLifting optimization plan, proposed optimization algorithm:
  Given the traverse head and connectivity function, the traversal method is specified
• Researched the principle of self-executing JAR

• Rate-limited compilation
• Custom rate limiting