ASU Meeting Agenda

June 4

• TCRS'24 Paper --> https://github.com/asu-kim/asu-kim-pubs/issues/16

May 3 / 10 / 17 / 23 / 28

• Arrange two meetings every week with @the-systematic-chaos --> Sharing through WhatsApp on Tues/Thursday

• Research -> TCRS'24

  • Paper writing --> Title: Edge-based approach for time-sensitive applications with resource-constrained robots (tentative)
    • Initial latex file with TCRS format --> https://github.com/asu-kim/asu-kim-pubs/pull/10
    • Dig up more related works --> Reviewing papers of IEEE Embedded Systems Letters (Done)
      • Edge computing, Computation offloading, Computer vision, Real-time ML (inference) (e.g., AnyNet)
      • Robot control, Resource-constrained (no camera) robot

• Experiment --> Collision Detection implementation

  • Order a new robot and a Raspberry PI 4 --> Will arrive 5/22 at 10 p.m.
  • RPI4 and SunFounder SVR (Smart Video Robot) assembly --> Start by @the-systematic-chaos from May 23.
  • Implement two same programs in Python and C
  • Check the performance of YOLOv5 LF example in our workstation --> 25 FPS
    vs C ver. --> CPU + YOLOv7-tiny: 560ms in Laptop, 307ms in Workstation
    vs C ver. --> GPU + YOLOv7-tiny: 30ms in Workstation (after updating Cuda 12.4 + installing cuDNN)
    Execution log on our Lab workstation - 30ms
    Object detection results

• rust-rti maintenance

  • #54
  • Synchronize the latest reactor-c code

Apr. 12 / 19 / 24

• Research -> TCRS'24
  • Experiments
    • Wi-Fi connection in Pololu --> Obtain a new USB connector to connect to the WixelConfig program
      -Testing the current serial wireless sdk on the module. It seems that there is already a TX and RX mechanism. Need to make sure messages can be TX-ed. --> The plan is changed to use RPI4
    • PololuDriving.lf implementation
      • Check cython (Write Python First and then compile it into C)
      • Collision detection --> Analyze the output results of object detection (in progress)
      • Camera reactor --> Testing v4l2 Linux src with the purchased webcam --> Obtained an image as a .ppm file
      • ppm to bmp conversion --> Apply netbmp library --> https://github.com/asu-kim/asu-kim-pubs/pull/11
• Review/Approval request
  • https://github.com/lf-lang/rust-rti/pull/50 (65.75%)
  • https://github.com/asu-kim/asu-kim-pubs/pull/2
  • https://github.com/asu-kim/asu-kim-pubs/pull/7

Apr. 5

• Research -> TCRS'24
  • Experiments
    • Wi-Fi connection in Pololu --> Solderings by Prof. Kim
    • PololuDriving.lf initial implementation
      • Object detection with Darknet: 7 seconds/image with CPU, 0.4 seconds/image with GPU in Workstation --> Purchase a webcam
      • Camera reactor --> Analyze NNStreamer Cam example
• Unit test implementation (server.rs) --> https://github.com/lf-lang/rust-rti/pull/50 (65.75%)
• Integration tests progress: (62/87) federated tests --> Pending

Mar. 15 / 22 / 29

• Wi-Fi connection in Pololu --> Purchase New USB Wireless Module
• Research --> Initiate TCRS'24 preparation
• Unit test implementation (server.rs) --> https://github.com/lf-lang/rust-rti/pull/50 (In progress)
• New feature/Integration tests progress: Clock Synchronization (Done) + (62/87) federated tests

Feb. 28

• Wi-Fi connection in Pololu
• Unit test implementation (server.rs)
• Tracing implementation for the analysis of federated examples
• [Review&Approval] Tracing
• [Review&Approval] Address Advertising/Querying
• [Review&Approval] RwLock instead of Mutex
• Research topic searching - Real application with real devices network + Fast detection with LF + performance + ...

Feb. 23

• Request to PR approvals (CI for integration tests with LC federated tests)
• Tracing implementation for the analysis of federated examples --> Done
• Pin-set headers/Soldering Kit/Multi-meter preparation
• Integration test PR in lf-lang/lingua-franca --> Pending (Will update after passing all federated tests)

This is a historical meeting agenda rather than a single actionable task. Review the unchecked items and linked entry points, including PololuDriving.lf, server.rs, the federated tests, and the Raspberry Pi experiment references. No single completion criterion or current starting point is defined; the agenda would need to be narrowed before work can begin.