Status Report 7(11.3-11.9)

Team

No changes made

Risks have been solved, the new controller board arrived

Schedule on time

Ruohai Ge

  1. Gather more data for the yolov3 ball training ->  800 self-labeled pictures
  3. Integrate the vision pipeline with the new learning inference.
  4. Set up the physical environment for the camera
  6. Schedule: On-time
  7. Deliverable for next week
    1. Fine-tune the vision part to make sure it can predict actual real-world designated point when a ball fly to that point
    2. Help Zheng to do the camera transform
    3. Write Bluetooth pipeline to connect the Vision and Hardware module

Xingsheng Wang

  1. Assembled the whole mobile platform system with the trash can
  2. Finished coding for the PID control of the car
  3. The controller board has arrived, so next week I can test out the whole platform system.
  4. Schedule: on-time
  5. Next week: A platform that can move to the location transmitted through bluetooth within 0.5s.

Zheng Xu

  1. Implemented real time inference for yolo3 detection inference.
  2. Integrate vision pipeline with yolo3 detection inference.
  3. Tested real time speed and accuracy of new pipeline. The new pipeline has relatively the same latency as the old pipeline. Plus, it has about the same false negative rate and much lower false positive rate.
  4. Help setup initial physical testing environment.
  5. Schedule: on-time
  6. Next week goals
    1. Implement camera coordinate to world coordinate transform.
    2. Help build final physical testing environment.
    3. Merge vision pipeline with mobile platform.