This week, several students and I went to go visit the Tesla Gigafactory for a tour. One of the REU members organized the tour and it was an interesting experience seeing what goes behind the scenes of a top scale company. I was impressed on their high reliance of robotics when manufacturing. Large industrial arms with a preset program did the same task over and over which increased the production speed. They also have a robot that autonomously navigates across the work floor with some sort of laser scanner. This factory mainly focuses on batteries and the tour guide shared the insights on what it is like to work for Tesla. Overall he said that people usually only last 2-3 years as it is a tough place to work due to the speed.
I also met up with Yu Hang this week and we caught up on what has been happening in the lab and in Reno. I hope his internship with Tesla goes well.
This week was spent working on the transform code to get the human position in the odometry frame of the robot. The robot navigates based off this frame, therefore it is essential to have the human position relative to it. It took a lot of debugging, yet I got it to work. Currently it only works for one person though. I also created a new map of the building using the Pioneer because the map I got from the professor was made on the PR2 and the Pioneer was having difficulties navigating through it. The map was made using gmapping. I also conducted the experiments in order to have data on the poster and the paper. One scenario consisted of a static human and an obstacle also in the way. The socially aware navigation planner differentiated between the human and obstacle and based its future trajectory on this classification. I also got the YOLO camera data to show on RVIZ. The distance of the located bounding boxes is shown with a marker and this marker can undergo a similar transformation as the human in order to have the robot localize these objects. A picture of this is shown below. The green dot is the frame while the blue rod is me being detected with the leg detector package. Next week is my last week and it will consist of turning in all the final deliverables. A poster presentation will take place on Thursday and the final paper will be due on Friday, so that will what I will be working on. Yet due to the printing process, the poster was due on Monday so that is another thing I completed this weekend.
I learned that ROS is a powerful tool in robotics and I should continue studying it to get better day by day. If it wasn't for this experience, I don't think I would know what I know now as I rarely used ROS in the lab. It is a difficult concept to grasp, yet with practice, it can be learned well. I also went to a graduate school workshop that the REU hosted and realized that I need to start doing research on what I want to do after my undergraduate. I still have two years left, yet it is a lot of preparation. I don't know much about graduate school because nobody in my family has attended, so it was helpful having someone explain the admission process and how to choose a school.
This week I met Kelsey, an undergraduate REU student studying computer science at Ithica college. Her project consisted of a machine learning algorithm that had an image as an input and then determined the trustworthiness and attractiveness of a person.