This weekend was spent showing the non-Seoulite members of the cohort around Seoul. The sights that the cohort had chosen to visit Saturday included Chandeokgung Palace and Namsan Tower, which were chosen because neither the Seoulites nor non-Seoulites visited them.
The Secret Garden in Chandeokgung Palace.
The next day, the Seoulites among the cohort showed the others around Itaewon, the foreigners’ district of Seoul, meeting for a rather heavy brunch at Bada-Sikdang.
A picture of brunch that day, which was some kind of Budae Jjigae. While this was indeed tasty, the author maintains that this is not an appropriate choice for brunch, as it is too heavy for such a meal.
After the meal, the group discussed what to do next. Kervin, turning out to be an audiophile, then suggested to visit a Hyundai-exclusive music shop called Plastic and Vinyl. As the Matt and Nathan had to leave for Daejeon (and Kervin had to leave to do some barhopping with some friends at SNU), the group separated at 2:00. Now separated from the rest of the cohort, Ryoma then went to the War Memorial of Korea, seeing as it was down the road.
An indoor vehicle collection in the War Memorial of Korea. The memorial also had an outdoor collection that had fighter and bomber jets, helicopters, missile platforms, and much more. The museum also satisfied the author’s itch to learn more about the Korean War, which was omitted from the National Museum.
Ryoma’s studies on URDF and Xacro files finally paid off when he finally got his testbed instrumented with the cameras he needed. Armed with the knowledge of Xacro macros, topics, and more he had command over his simulation environment at last.
Unfortunately, Ryoma is facing one problem in his software. While the implementation of the ROS packages were a success, the base software only simulates the D435 camera (as opposed to the D435i camera), making it inadequate for his needs. Specifically, the D435i has a kinematic sensor called an Inertial Measurement Unit (IMU) that measures the orientation and acceleration of the camera, and by extension the robot, to retrieve its location in space. While not strictly necessary for visual SLAM, such instrumentation does improve the performance of such algorithms greatly.
Ryoma has two options to overcome this. Either he takes the easy way and inserts a preexisting ROS IMU plugin and insert it into the packages, or dig directly into the code and create a new sensor from scratch.
Ryoma does not like talking about himself, hence the relative brevity of his posts. Regardless, Dr. Moser and his workshop sessions forced him do do so, as it was time for the job talk section which, of course, involved elevator pitches. The current subject of the sessions naturally evoked Vietnam-style flashbacks about job hunting. That is not to say Ryoma is not thankful for these sessions- they are undoubtedly useful and will prepare him for stuff he will have to suffer through. He may even improve to the point where job hunting is not traumatic.
To improve his style, Ryoma dusted off and memorized an old elevator pitch he had and delivered it to the best of his ability. To his surprise, the others said it was quite competent and conversational, save for the ending which petered off. Either Dr. Moser and his cohort are being too nice to him, or the job junts weren’t as bad as Ryoma thought.