After nearly an academic year of work on our project, we have finally made it to the Final Design Review, or FDR. The event started out with a demonstration of our prototype to our engineering liaisons, Kyle Bush and Kevin Seymour, which went well. Unfortunately, a wire on the LiDAR was broken, which meant that we could not show it off at that time; however, we quickly soldered on a new connector at the venue to get the prototype functional. After this, we attended a keynote presentation on the experience of professional engineering, which was followed by a very polished project presentation from this year’s best team, Desigators. Next, teams went to their assigned rooms to present their projects to an audience of the other teams in the room and those teams’ liaisons and coaches. We were matched with Crystal Clear Waters (sponsored by Duke Energy) and SolarSafe (sponsored by NextEra Energy, who is the parent company of our sponsor, FPL); the room’s theme seemed to be utility companies. Our team’s presentation went very well, and we stayed in the time constraint and were able to answer all questions. Finally, we returned to our prototype (see the image below), and the public was invited in to inspect the projects. With our LiDAR system and alert system working, our project had some interesting (and loud) interactive components. We ended off the event with a conversation with Kyle, who was very happy with the progress that we had made on our project.

We here at the RoadWatch team want to thank you for following our project over this past year, and wish you all well. We hope our experience can inspire future engineers to take IPPD and have similar experiences developing their own projects. It has been a pleasure updating you weekly with our progress. Go Gators!

With the project wrapping up, we need to showcase all of our work and progress to our sponsor, FPL. Therefore, one of our objectives this week was the drafting of the FDR (Final Design Review) Report and Presentation. Unlike the SLDR from last semester, whose objective was to provide a concise overview of the work performed, the expectations of the FDR report are to provide complete and total documentation of all progress, discoveries, tests, and failures (the FDR presentation still shares the SLDR goal of being information efficient). After completing the presentation and some practice, we partook in the FDR Peer Review event, along with the Splunkernauts, CyberSign, and Fantastic 5 teams. In this event, each team presented to the other teams, and received feedback forms from all audience members. This gives an opportunity to make any necessary adjustments before the presentation with real stakes. Also this week, substantial progress was made on the prototyping and integration front. Evidence of this can be seen below, where the assembled Detection and Alert systems are shown. With next week being the FDR, it will be our last post. Be sure to come back and read it!

The pinnacle event of IPPD, the Final Design Review (FDR) is just over horizon, on April 23rd. By this date, we not only have to complete our prototype, but also need to have finished our FDR Report. This is a massive document consisting of five volumes, which describes our project in such detail that it can be passed off to our sponsor to continue making progress on it. We have been gradually working on this document over the course of the semester, but with prototyping requirements waning we have started allocating more resources to the completion of this document. Fortunately, we have made progress, and are on track to have a first draft completed by next Tuesday. We have also made some prototyping progress this week, though it is mostly on the integration front. We have successfully assembled the alert subsystem. Progress was also made on computer vision, as a versioning issue with a GPU acceleration package was resolved. To see how our report evolves, be sure to read about in next week’s post.

This Tuesday (April 2nd), the Road Watch team took part in the second Prorotype Inspection Day (PID2) Event. Here, like with the same event last semester, each team presents their progress on their project prototype to several judges with sufficient background to assess the deliverable. It is very comparable to a science fair in this regard, but features a greater emphasis on engineering output than on research results. Our prototype was not where we had hoped it would be, due to some last-minute error encountered during the integration stage. We ran into an issue where the LiDAR would not work within its acrylic enclosure, which we later discovered was due to partial reflection off the acrylic. We also were not able to use the MIPI CSI cameras with YOLOv8 on the Jetson Nano, because of some driver issues. These combined to produce a non-integrated prototype with some band-aid solutions, so we fell back on presenting the functionality of each individual subsystem, which ended up going well. Despite this, our judges generally gave our project a good ranking, with most either awarding Excellent or Good marks in the rubric categories. We certainly have a lot of work to do before the Final Design Review, or FDR, especially to overcome these integration hurdles. Stay tuned to see how that pans out.

As has been alluded to extensively in previous posts, Prototype Inspection Day (PID) is right on the horizon; in fact, it is next Tuesday (4/2/24). Therefore, in class this week, we prepared our strategy for the day, and shared the strategy with the class. Our strategy generally breaks down demonstration responsibilities based on project area. First, Rolando will open with a quick elevator pitch, to acclimate the judges with the purpose of our project. Then, Darrion offer a demonstration of LiDAR functionality. Following this, Skyler will give an overview of the mechanical considerations of the project. Then, Evan will demonstrate his implementation of the predictive algorithm. Will will then show the camera vision project aspects, and discuss testing efforts. Rolando will then give an overview of the system interface, and demonstrate how users may use the system. Finally, Billy will discuss administrative considerations associated with the project. The class generally approved of this approach, made some recommendations, including augmenting our demonstration with a graphic of the system architecture. Be sure to come back next week to see how PID went.

With PID looming one week away, the Road Watch team has been rapidly working on getting every part of the system working. Fortunately, we have some good progress to report. First, we successfully developed our external shell, and fabricated the internal spacers. Next, our research concluded that a LSTM (Long-Term Short-Term) network would not be a viable approach for vehicle position prediction (that’s a shame, but discovering such things is important). However, some initial testing indicated that the YOLOv8 model could be run successfully on the Jetson Nano, which is a critical state in our camera vision pipeline. Finally, we successfully assembled a second revision of the Detection PCB and a first revision of the Alert PCB (pictured below). Come back next week for more progress.

Although it is spring break, and despite the note last week, some progress was made on the progress this week, which the Road Watch team would like to share. Firstly, we believe we have solved our issue with the 1D to 2D LiDAR conversion. Some testing (see the first image below) indicates that the mirror used for the conversion was not large enough, and because the LiDAR we purchased had its transmitter (i.e. laser) and detector off-center, this would lead to some of the beam not being reflected depending on the angle of the mirror. Therefore, we are confident that this issue can be resolved by increasing the mirror size, which we will test next week. Additionally, revision 2 of the Detection Subsystem PCB and revision 1 of the Alert Subsystem PCB were completed and ordered, and arrived this week (see the second image below). The former was redesigned to address some issues and deficiencies of the first iteration, while the latter is needed for Prototype Inspection Day (PID) to show a complete system. As PID approaches, be sure to come back weekly to monitor our prgress.

With the feedback from QRB2 in mind, and the Prototype Inspection Day (PID) rapidly approaching, there are only so many weeks left (two to be exact, plus spring break) before we have to present a fully integrated and working model of our system. Therefore, we have no time to waste on project development. Fortunately, we have some progress to report for this week. Substantial progress was made this week on the electronics of both the Detection and Alert systems; the Detection PCB was revamped to correct previous errors and to add new features, and the Alert PCB nears completion. They will hopefully both be ordered this weekend. Additionally, progress was made on our stereo vision pipeline, and some models were identified which could perform some critical stages. Prototyping was successfully performed on the shell of the Detection system enclosure (see the figure below for a model of what this looks like). Finally, a technique for predicting the future position of vehicles with constant position was successfully developed. Next week is the UF Spring Break, so unfortunately there won’t be a new post. However, check back in the following week to see how our project has further progressed.

This week, the Road Watch team completed its second Qualification Review, QRB2. As explained in last week’s post, the QRBs are an event where teams receive constructive feedback and a risk assessment from a group of experienced faculty. This time, our panel consisted of Dr. Rui Guo (our coach), Dr. Vincent Tocco (the MaeChem Silanes coach), and Dr. Eban Bean (the Crystal Clear Waters coach). The panel seemed optimistic about our project, but identified some issues. They identified that, despite the large amount of progress that has been made, our prototype has a lot of work to go before it is an MVP (minimum viable product). In particular, they were concerned about our alert subsystem, for which no progress has yet been made. Additionally, they pointed out that it was not currently clear that our TPMs (technical performance measures) would be satisfied. Lastly, they were again concerned about the accuracy of the system’s predictive abilities. Their recommendations focused on testing; one panelist provided a location in Gainesville where vehicles merged together as a good spot for testing. Therefore, the next round of designing will focus on completing the hardware MVP, to enable the recommended testing and to prove our system’s predictive capabilities. Next week will feature more prototyping, so be sure to come back next week to keep up with our progress.

This week, the Road Watch got ready for its second Qualification Review Board, QRB2. As discussed previously, the QRB is an event where we present our project’s progress to a panel of expert faculty and receive feedback on our project. In particular, the presentation of and feedback on project risks are important, as these will dictate the success and failure of the project. Fortunately, after these weeks of work, we have plenty of meaningful progress to show the panel. For instance, we have made substantial progress on a fully-integrated prototype of our detection subsystem (see the image below) and are eager to share this. Additionally, we have developed a new architecture for the camera vision algorithm, which we are interested in getting feedback on. We also have some anticipated risks to share, such as our uncertainty about our system’s predictive capability and the potential for thermal regulation issues. Be sure to follow up next week to read about the results of the QRB2 event.