November 2020
tl;dr: Very thorough description of using HD map for TFL recognition.
Overall impression
Although the perception models the paper uses is quite outdated, it has a very clear discussion regarding how to use HD maps online.
Also refer to TFL map building with lidar for a similar discussion.
Key ideas
- Prior maps (with lat/long/height of TFLs) improves accuracy of recognition and reduces algorithm complexity.
- Task trigger: Recognition algorithms do not have to operate continuously as perception begins only when the distance tot he facing TLF is within a certain threshold
- ROI extraction: this limits the search area in an image
- Estimate the size of a TL
- Procedure
- RoI extraction with safety margin. Slanted slope compensation. Road pitch needs to be stored in the HD map as well.
- Detection locates TFL in image
- Classify state of TFL
- Tracking estimate position of TFL. Threshold for association should adjust based on distance.
Technical details
- The effect of pitch (on a bumpy road) is bigger for TFL at long distances. On average the pitch change could be up to +/- 2 deg.
Notes
- Questions and notes on how to improve/revise the current work