“An interesting day ahead”, I smiled and said to myself. I could barely hold my excitement and ran to finish my shower. I headed to the factory along with several other participants from all over the world. That’s one of the best perks of a global hackathon, you get to meet amazing people with diverse backgrounds. I helped the cab driver in giving directions as I knew the route to company from previous day’s trip. We reached the factory and the organizers from PIX briefed us on the hackathon and gave a tour of the factory, arena and competition rules. We had to finish 7 missions, where each carried 10 points and 30 points for the time taken to complete. We got back to the factory after looking at test site with huge and excellent roads within tech park zone. (Guiyang has great roads in general, as well)
We continued our day with the introduction session of all the participants. Eduardo from Spain who had attended the hackathon last year as well, took up the task of Minerva McGonagall. We formed three teams consisting of approximately 5–6 people each. The team captains were decided — Samuel (China), Dr. Kerem Par (Turkey), Sam Pfeiffer (Spain) based on experience.
My team :
Dr. Kerem Par — CTO of Adastec from Turkey. Known for his calm mind, he had Autoware (open source self-driving software) guide completely printed with spiral binding and an RQT graph of the entire stack.
Mustafa Erdem — Adastec from Turkey. The most excited and enthusiastic member of the team, his age is just a number. Known for his belief in uncertainty, his statements possessed high entropy, “Maybe, maybe not, I don’t know.”
Italojs — Computer vision engineer from Brazil. He saved my entire trip by lending me an adapter for my laptop which I had conveniently forgotten to carry. Known for his chill personality, he worked on many ad-hoc aspects of the missions.
Dr. Ali Ufuk Peker — CEO of Adastec from Turkey. Known for his witty comments (which you will read soon), he was the “mapper” of our team. The most used words by him during the competition : “vector map”, “stop line”, “dhur dhur” (means stop in Turkish).
We decided to have lunch at 12 noon. Nancy Lee (Community Manager PIX, aka tourist guide, wonder-woman) led all of us to a cafeteria that is close to the factory. Life in China is quite harsh for a vegetarian as the options are highly limited with just plain rice and some vegetables, but that didn’t affect me much. I tried eating with chopsticks for the first time and ended up being the last one to finish out of the entire group. (I improved eventually, but Rohan would disagree)
We returned to the factory and pix engineers gave a short overview of hardware and handed over “a toy” to each team. (Toy is a CAR). Each car had an onboard computer, screens attached, 16 channel LiDAR, CAN bus, RGB camera, along with a huge monitor to work with. For an undergrad student who had limited access to state-of-the-art hardware resources, this was like a “playground” for me. That’s when I remembered Chase’s (COO of PIX) speech that morning, “We encourage innovation through freedom”. I liked the freedom and couldn’t ask for more.
I got into the car with lots of excitement and typed some keys to get the specifications of the beast server that was lying in the trunk. “Intel i7 with Nvidia RTX 2080”(released last year’s end). The screen’s display was off. We decided to spend the first day in setting up the server and calibrating the sensors. We listed the software requirements and had a brainstorming session to get a rudimentary plan for the missions. I installed necessary drivers, toolkits and set up the software needed (like Nvidia, CUDA, Tensorflow, ROS, Autoware etc.) The internet was extremely slow at the factory as we had to use VPN for unrestricted access. By the end of the day, we managed to get the PC running and decided to test LiDAR. But things didn’t go as per our plan. We configured the network properly, yet we were unable to get Velodyne data. The red lines on the terminal stared at me :
“DriverNodelet::devicePoll — Failed to poll device.”
I tried several ways to debug it by running different versions of the driver and all the efforts went in vain. We decided to switch to laptop which also had Autoware installed. “Login. Launch Autoware. Launch Velodyne driver. Baaam!!” We were getting the data.
For a moment, all of us were elated. But the point cloud data did not look convincing. I walked near the LiDAR of my car and realised that the data we were receiving was not from our car. Ali walked around and waved near each car to find out the source. Sam’s team was broadcasting the LiDAR data and since we were on same network, we were receiving the data as well. (This was later rectified). We sat back with disappointment and asked for a new LiDAR only to get the same errors again. By this time, the rest of my team was tired and we had to go back to the hotel.
I barely got sleep that night. I couldn’t stop thinking about what might have caused the errors and how I could fix. Unfortunately, the usual sites like Google were blocked in the hotel which made me helpless but wait for the next day. Another sunrise, with a restless mind and hopeful heart, I made a rough agenda for the day and possible fixes for the driver. The last fix was quite painful and that was my final resort — “A complete and clean installation of Ubuntu” again. It seemed daunting but I knew it would work for sure.
“Today, I need to fight with the LiDAR”, I told, as I sprinted to the reception to take a cab.