2020-06-01

This morning, as soon as I arrived, I went to analyze the issue with Yiming. We spent the entire morning on it. The result was another problem caused by a previously known issue (point loss due to differences in the old compiler): when capturing the road, it happened to be near the missing segment, so the lane offset turned out to be a negative value.

I ended up working until around 4 AM, after a period of inefficient and painful late-night work (it seems that under the current state, whenever I stay up late, it almost always leads to inefficiency).