12/7/24
This week I orchestrated the majority of the final presentation slides and content. After that, I practiced the presentation and presented to the class on Monday this week. On Tuesday, I picked up materials to build the track for our demo and worked on debugging the radar on the Jetson. I worked with Ankit to try to set up the GPS, and we ran into a lot of issues but discovered after some time and coordination with group C6 that we have to run the code on a Linux system or it will not work. On Wednesday, we had presentations again, and I helped Ankit reorganize our materials box and set up the traffic lights and dump truck for our demo. On Thursday, our team was able to get a working but not robust situation to SSHing into the Jetson, mounting our system onto our RC car, and getting some video for testing of the RC car driving on our track. On Friday, I dove deep into radar testing, and I implemented a smoothing algorithm using sliding window averaging to help calm down the noise in the radar readings. I also did accuracy testing and brainstormed metrics to quantify the performance of the radar system.
My progress is currently on track.
In the next week, I hope to test the radar again both inside and outside and obtain radar output graphs, variance measurements of different sliding window averaging sizes, and accuracy measurements for both indoor testing and outdoor car testing. I also hope to develop the content for the poster with Ankit and gather car testing video for forward car departure detection. Lastly, I will be demoing with my team on Friday and finishing off strong with the final report due on Saturday, but we are hoping to get this done before Friday because Ankit and Eunice both are traveling home on Saturday.
11/30/24
This week I organized a Final Stretch checklist that needs to be accomplished before the semester ends so we can wrap up our project successfully. Our team has been experiencing an increased amount of stress, so I wanted to lay out exactly all of the building blocks that are left for us to complete before the end of the semester to chunk the larger problem of finishing our capstone project into smaller digestible pieces. As part of the checklist, I helped to order the last parts we will need to complete the project with Ankit and Eunice: remote-controlled traffic lights, an RC car large enough to hold our product, and a battery pack for our Jetson so we can run it without having to connect it to an outlet power source.
I am currently on schedule.
Tomorrow, I will gather the materials to build our track and finish the Final Presentation slides and rehearse since I will be presenting for my team this time around. Aside from that, I want to graph the noise from the radar sensor and analyze the pattern to see if it’s predominantly lower noise, higher noise, or normally distributed noise. From this, I can devise a more educated plan to reduce the radar’s noise. I also want to build the demo track and mount everything onto the RC car with Ankit and Eunice. After that, as a team, we want to rigorously test our system in a demo setting by the end of this week.
11/16/24
This week I helped to debug the forward car departure. We were having a lot of trouble figuring out what 3D range of space the radar was detecting in, so I communicated with the sales associate for the radar detector that we bought, and she gave us some really great insight about how to use the API more effectively to filter out noise and where in 3D space the radar actually detects. Ankit and I filtered out noise coming below 1m readings on the radar detector. After doing that, we did the testing described in the team report in 1207 of walking away from the detector and seeing whether our boolean forward_car_departed was set to true or not. Currently, I think it’s setting correctly, but it isn’t resetting well, so I will be working on that. Additionally, we want to find a different more open testing space to avoid as many extraneous noise readings as possible. I also did a lot of planning for powering our Jetson without using an outlet. I helped to make a plan for powering it while our device is in a car and powering it while our device is attached to an RC car. We decided we can use the car smoker to power the Jetson in a car (and all of the other devices because all of the other devices are powered using the Jetson). We are still working out the exact details for powering the Jetson with the RC car, but we’ve found a DC power source that looks promising. We just need to do a little more digging to confirm that all of our parts will be compatible with this. I also led communication with team C6 to organize the type of antenna that we are going to buy, and I put in an order for that this week. Lastly, I drove around to help collect video data for us to test LD and TLD with. Ankit and I are debugging LD specifically with this video data that we gathered, and we have isolated that the problem with our annotations is that they’re picking up a lot of noise. We’ve been discussing how we can mitigate this moving forward, and we are figuring out what particular ways we want to filter the annotations to isolate the ones on the actual lane markings.
My progress is on schedule this week for debugging FCD, LD, and TLD, and it’s also on progress for organizing parts for taking our device from the computer lab to an actual car or RC car.
I’m hoping to finish debugging LD and TLD with my team early next week for interim demo, plan and have a great interim demo on Monday and Wednesday, and make significant progress to debugging FCD.
11/9/24
This week I helped Eunice flesh out the necessary steps for developing lane detection and traffic light detection. We discussed the balance we wanted to have to implement these two features with YOLOv8 and OpenCV, and we initially decided to train a YOLOv8 model on a custom traffic light dataset and then use that output to feed into OpenCV code to detect traffic light changes. I worked on training the YOLOv8 model, but many things were not going well. I had a really hard time trying to open our initial dataset due to the way it was zipped, and after that had already taken too long I was able to find a separate dataset. I tried to train that one using Google Colab, but it took so long that I got kicked off of the GPU I was training on (it ran through the night). I tried training it again with half the epochs but ran into the same issue, and this just took me a super long time. After those failed attempts, our team decided to try to implement traffic light detection with Hough circles, but keep this training in the background in case we could figure it out later.
Later in the week, I was working with Ankit on the forward car departure feature. I wrote the initial algorithm in a file that Ankit had already written to draw data from the radar and the camera. We spent the day today trying to debug, and we formulated a good list of the next steps for continuing to dig into the development process for this feature.
My progress on average is currently on schedule because I’m behind on traffic light detection with Eunice but ahead with Ankit doing forward car departure.
I’m hoping to finish traffic light detection and forward car departure by the end of this week as well as organize power sources and our startup + code script so we can run our product without setup or having to plug things into a wall outlet.
11/2/24
This week I helped to debug the Jetson Orin problems for some time, and then after we figured out that our camera was actually not viable, I helped work with Eunice to select a new camera that we could order after returning our current camera. We also were able to pick out a GPS from the available inventory since our new camera does not have a built-in GPS. Additionally, I organized our schedule again as reflected in our GANTT chart and more detailed development schedule to account for a second week of being unable to develop due to failing to debug our Jetson Orin problems.
Since we had so much trouble with our Jetson Orin and we realized our camera wasn’t viable this week, my progress is behind. To make up for this as I mentioned I reorganized our development schedule, I will be completing data synchronization and LD + TLD with my team and Eunice respectively next week to make up for this lost time.
By the end of the week next week, I hope to have finished LD + TLD with Eunice and wrap up the week with unit testing.
10/26/2024
This week I helped my team set up data transfer from the radar sensor to a computer and debug the Jetson since it’s been failing. I also found a good resource for running a program upon Jetson startup so later when we go to test this we can automatically start running our program upon turning our Jetson on. I also did the individual ethics assignment.
My progress is behind because we haven’t been able to get the Jetson up and running. I was supposed to be implementing the forward car departure warning this week, but I have to push it back to next week or until our Jetson is working. I will be grinding on getting that feature out as soon as our Jetson is up.
Next week I am hoping to help my team in getting our Jetson up and running. Then, I’ll help my team complete data extraction and synchronization. Lastly, I will finish the leaving car detection.
10/20/2024
This week I helped my team to finish the Design Report. I helped to delegate different sections to our team members, and I wrote my individually assigned sections: I wrote the Abstract, Index Terms, Introduction, Test, Verification, & Validation, Project Management, and Related Work sections as my individual contributions to this assignment. I also talked with Ankit and Eunice to help them set up the Jetson Orin and the dash cam.
My progress is currently on track.
Next week, I hope to help Ankit and Eunice set up the Jetson Orin and the dash cam as well as work on extracting data from our radar sensor. I also am planning on implementing the majority of forward car departure detection after the radar sensor setup is complete. By the end of the week, I also will sync with Ankit and Eunice to discuss my implementation so far and do some testing with them.
10/5/2024
This week I helped delegate work so our team could collaborate effectively to create the Design Presentation slides. After delegating this work, I specifically wrote one of the quantitative design slides, the test, verification, and validation, and the project management slides. I fleshed out a lot of our testing plan while creating this presentation and spearheaded the conversation surrounding defining the exact measurements that will define our integration testing infrastructure. I also helped initiate communication within our team to descope the forward collision alert officially from our integration testing – realistically it is too dangerous to human life and to our testing system to include this in integration testing, so the plan for now is to vigorously unit test.
My progress is slightly behind. I haven’t started working on the Design Report yet; however, I will have the time to grind on it tomorrow.
I will complete my portion of the Design Report by the due date next week, and I also will help make sure that all of our equipment is ordered by the end of next week so we can hit the ground running when fall break ends. Additionally, I want to help Ankit get the Jetson Orin working.
9/28/2024
Throughout this week I worked with my team to help flesh out the exact pieces of hardware + software libraries that we will use for implementation. We individually did research particularly on the dash cam and the distance sensor that we are ordering. I coordinated with a sales representative to discuss details about the radar sensor we are planning on using to clarify API use and radar specs, and I also reached out to the robotics department to check whether or not we are able to borrow any sensors from them (with no luck :/). I also set up the skeleton for the design presentation slide deck and helped delegate individual work so all of us can pull this presentation together by Sunday night. Lastly, I created the development schedule under another sheet under our GANTT chart spreadsheet as well as updated our GANTT chart development section to reflect the changes.
My progress is on schedule.
Over the next week, I’m going to finish developing my assigned slides for the design presentation, present the design presentation, and give feedback on other student’s presentations. Additionally, I would like to make significant progress towards finishing the design review by EOW next week.
9/21/2024
This past week I was able to contribute to making the proposal presentation slides by working with my teammates requirement by requirement to develop the content of the presentation. After the content was created, I selected a theme and design for each of the slides and talked with my teammates about which text was most important to put on the slides as well as what images could better help users understand the presentation’s content. I also read the Design Presentation guide and brainstormed ideas so I could set myself up to contribute as best as I could for when my team sits down to develop it.
My progress is on track for this week. This upcoming week, I hope to select a theme and design to display the information well for our design presentation. I feel we could have done better in this respect for our proposal presentation, so I’m hoping to find a really nice display of information for this next presentation. I’m hoping to finish our design presentation slides by EOD Wednesday. Additionally, I’m looking to finalize the Lidar sensor we are going to use, order it, and learn about its connectivity interface to begin having ideas for implementing forward collision warnings and forward car departure.