watonomous.github.io

1. Knowledge Base
2. Knowledge Base Home
3. Management Tools & Training
4. End of Cycle Reports
5. W20 End of Cycle Reports
6. W20.1 Software Group EoCs

[ Knowledge Base : W20.1 Perception EOC ]

Created by [ Anita Hu], last modified on Mar 01, 2020

Link to EOC guide

Heading

Name: Anita Hu

Date: February 13, 2020

Development Cycle: 1

Division: Software 

Group: Perception

Status

[GREEN]

OKR Timeline Progression

[W20.O1] Roadline Detection

0% → 30%

Good progress has been made towards the tasks.

• Stop line detection is in testing phase, looking into testing with ros bags
• Found a fast repository for depth map generation, currently converting it to a ros node
• Roadline binary mask is almost ready to go, just need the BEV transform

[W20.O2] Traffic Light Recognition

[W20.O2.KR2] Traffic Light Position

0% → 20%

Good progress has been made towards this task.

[W20.O3] Multi-Camera Detection

0% → 10%

• New camera lens arrived
• Waiting for PoE to arrive and then we can have all three cameras running
• Tested inference time of running multiple obstacle detection nodes vs doing detection on a batch of images. The latter works a lot better. We can get 7 FPS while running detection on 3 cameras. Perception pipeline diagram will be updated. 

[W20.O6] FPGA CPU Optimization

FPGA is no longer supported. Yifan updated the OpenVINO to the newest version. 

The next step is to look into using VTune for CPU optimization.

\

Link To Retrospective

\

Core Member Engagement

Group Engagement Rate: (11 - (1 + 0)) / 11 = 91%

Link to engagement level guideline

+——–+————–+———————————————-+ | Name | **Engagement | **Activity Summary | | ** | Level** | | +——–+————–+———————————————-+ | David | Meet | Multi-lane detection | | Wang | | | +——–+————–+———————————————-+ | Shusil | Meet | Road line binary mask | | Shapko | | | | ta | | | +——–+————–+———————————————-+ | Celene | Warn | Stop line extraction | | Chen | | | +——–+————–+———————————————-+ | Aswin | Meet | Traffic light position, arrow signs | | Visva | | | +——–+————–+———————————————-+ | Mojtak | Meet | Depth map | | a | | | | Valipo | | | | ur | | | +——–+————–+———————————————-+ | Janaka | Meet | Depth map  | | n | | | | Chandr | | | | akumar | | | +——–+————–+———————————————-+ | Feilan | Meet | Multi-lane detection | | Jiang | | | +——–+————–+———————————————-+ | Allan | Meet | Stop line extraction | | Zhao | | | +——–+————–+———————————————-+ | Sabrin | Meet | Road line binary mask | | a | | | | Luo | | | +——–+————–+———————————————-+ | Sincla | Meet | Semantic segmentation | | ir | | | | Hudson | | | +——–+————–+———————————————-+ | Martin | Meet | Traffic light position, arrow signs | | Ethier | | | +——–+————–+———————————————-+

Document generated by Confluence on Nov 19, 2021 12:21

Atlassian