This week I focused on system testing for our Focus Field prototype, specifically evaluating our core performance requirements using the fully assembled setup. For target lock accuracy, I conducted 67 human trials with 1–4 users in view and recorded system performance, resulting in a 77.6% success rate (52/67), which is currently below our ≥90% requirement. I also ran steering error tests by physically measuring angular offsets using a protractor on a flat plane at 1 meter, aligning with our system assumption of horizontal-only (2D) measurement. These tests helped validate our testing methodology while also identifying areas where system performance needs improvement. In addition, I continued working with our testing infrastructure and debugging, including identifying and fixing an issue in one of our test files related to automated evaluation. Overall, this week was focused on collecting real performance data and validating our testing pipeline.
Progress may be slightly behind in terms of meeting performance requirements, particularly with target lock accuracy currently below our threshold. However, this is expected at this stage since we are now transitioning from setup into real testing and identifying system limitations. The next step is to analyze the causes of the lower accuracy and refine our approach, whether through improved alignment, algorithm adjustments, or system tuning. Since our testing framework is now working, we are in a good position to iterate quickly and improve results.
Next week I plan to run additional system tests for interferer suppression and latency using the testing suites that have already been implemented. I also plan to begin intelligibility testing using our defined approach with speech-to-text comparison to measure ASR error reduction. The goal is to collect complete performance data across all major requirements and continue refining the system based on those results.