The most significant risks is the integration, which during our test, we found that sometimes the voice command was severely delayed due to unknown reasons, which only after several voice prompts, the corresponding output can be delivered. The current solution has been using multithreading to manage all long-delay tasks such as speaking the obstacle. The contingency plans including merge different voice prompt into one, spawn less process, and prioritize for high-important obstacles such as humans (since they are moving).
No change is made to the system design. No updates to the schedule. We are on track now for testing.
