Status Report 9(11.17-11.23)

Team

  1. Tested whole pipeline of the system and gathered some metrics for transform accuracy and position estimation.
  2. Major risk for the project: The pipeline is not fast enough right now. The platform would reach the target position approximately half a second slower than we wanted. Viable solution would be speed up kalman filter output sacrificing some accuracy and give some heuristic movement before the rough location of the ball is decided.

Ruohai Ge

  1. Finish taping the inner moving 1-meter surface reducing the drift of the mobile car
  3. Help Xingsheng tune the PID and weight distribution of the mobile car
  4. Finish the transform from camera frame to real-world frame
  5. Fine-tune and test the vision pipeline with Zheng
  6. Schedule: On-time
  7. Deliverable for next week:
    1. Integrate the vision pipeline with the mobile cart
    2. Write and test the entire pipeline
    3. Do the validation and metric stated in the design review
    4. Write the final report
    5. Prepare for the final presentation

Xingsheng Wang

  1. The mobile platform is able to move smoothly in straight lines, but the diagonal movements have a lot of drift, which reduces accuracy of the platform motion. I used tape to cover the area that the platform would cover, which reduces drift. However, the accuracy is still not optimal. I am tuning the parameters of the motors and also tuning the weight distribution to improve accuracy.
  2. I have also finished the final mounting of the whole platform. The battery is beneath the platform with double-sided tape to secure it. The trash can is taped down onto the elevated wooden platform.
  3. Schedule: on-time.
  4. Next week:
    1. Better accuracy of the mobile platform motion.
    2. Metrics and validation done.
    3. Write the final report and start preparing final presentation slides.

Zheng Xu

  1. Finished transform from camera frame to the real-world frame.
  2. Tested transform accuracy.
  3. Improved multiple features as detecting ball movement, ignore non parabolic trajectories and landing position check.
  4. Integrated transform and bluetooth to the general pipeline, the entire pipeline is now complete.
  5. Tested whole pipeline speed and accuracy.
  6. Schedule: On-time
  7. Next week:
    1. Improve pipeline speed.
    2. Get better transform parameters.
    3. Improve trajectory estimation output latency and accuracy.