What are the most significant risks that could jeopardize the success of the project? How are these risks being managed? What contingency plans are ready?
1) The accuracy of plate recognition is the biggest challenge for our project right now. The resolution of pictures taken by our camera is not high enough for OCR OpenCV algorithms. In addition, there are letters other than plate number (e.g. state name) that add noise to the model. We’ll look into more ML algorithms and consider cropping out the plate region as preprocessing.
2) We foresee that there may be communication latency. The designated parking spot may not be sent to raspberry pi’s at intersections in time, and drivers will need to wait for instructions for a longer time. We’re still thinking about solution to this issue.
Proposed solutions:
Proposed backup for (1):
We can move the work of recognition to the AWS EC2 servers which can sustain a higher processing load without too much latency; this will require the camera image be uploaded to the EC2 servers at a reasonable latency and speed; the current tested round trip time is 0.5 second on average under very moderate traffic; we have yet to test the round trip time under heavy traffic. This part of research is assigned to ruizhezh;
Proposed backup for (2):
The proposing of backup solution for this issue relies on solving (1) first; currently no proposals are concretely stated for this issue, but many are planned; ruizhezh will investigate further on this in the coming weeks.
Were any changes made to the existing design of the system (requirements, block diagram, system spec, etc)? Why was this change necessary, what costs does the change incur, and how will these costs be mitigated going forward?
No change.
This is also the place to put some photos of your progress or to brag about a component you got working.
Server API framework:
https://github.com/Roger-ZRZ/SR_18500_Spring2022/tree/main/ParkingAdmin
(the project is currently private, please request access and I will add you)