After the Teknofest 2021
What happened in the robotaxi — full-scale autonomous vehicle competition?
We’re happy to be with you again in our blog. In this blog, we’re writing everything about self-driving technology.
In this post, we want to talk about Teknofest 2021 robotaxi competition and celebrate the teams we’re sponsoring with their latest achievements.
Leo Drive supports university teams that have competed in autonomous competitions for several years. This tradition started with Yıldız Technical University’s AESK team competing in the Tübitak Efficiency Challenge — Autonomous Category back in 2018. The AESK team was the only contestant that could complete the track, and they enjoyed the 1st prize that year.
Since 2018, Teknofest has hosted many competitions (in 2021, Teknofest hosted 35 different competition categories with 13,000 contestants from 111 countries), one of them being the robotaxi competition where university teams build their vehicles, sensor kits, and autonomous driving software to meet and compete with their peers on the test track. We have seen the change of difficulty in the contest in passing years. And every year the sophistication of the competition is increasing.
This year in September, 24 university teams competed in parkour that consists of lane markings, real-world traffic lights and traffic signs, designated stops for picking up and dropping off passengers (not literally picking up passengers but imitating by safely stopping and waiting) and finally, a designated area for parking with allowed and not allowed parking spots.
Leo Drive sponsored Yıldız Technical University’s AESK team, Sakarya University’s SAITEM team and Istanbul Technical University’s İTÜ Solar Car team for the Robotaxi-Full Scale Autonomous Vehicle competition. The teams received LiDAR sensors to be implemented on the race vehicle. Students quickly succeeded in integrating the sensors onto their test cars. Also, they did a great job in AD stack development for the well-known four pillars of autonomous driving: perception, localization, planning and actuation control.
The AESK team (as the most experienced team in the competition) used many state-of-art techniques for Teknofest 2021. For example, they used a LiDAR-camera sensor fusion system to obtain better classification and distance measurement accuracy. They used state-of-art LiDAR-camera calibration techniques, which yielded very accurate results. They said,
First of all, cameras give us planar information, and we cannot determine how far objects are. However, thanks to our LiDAR sensor, we can get depth information from the environment. So, by fusing lidar and camera, we managed to detect objects and have their positions accurately related to the car.
The SAITEM team also used the LiDAR sensor for obtaining the distance of the objects in autonomous driving. They also mentioned LiDAR was particularly helpful in emergency braking algorithms. Besides that SAITEM team had a rigorous testing methodology. They said,
We had been testing our car for almost five months. We tried parking almost one hundred times.
Rigorous testing and repeatability are essential to the SAITEM team as they have outstanding experience in the simulation-based testing domain. This year, they joined the Shell Eco Marathon Autonomous Programming Competition and were awarded 3rd place out of 32 teams from 20 countries. All university teams we support have the ambition to join this competition as well. Last but not least, the SAITEM team also plans to build a level 4 autonomous vehicle apart from any competition.
The ITU Solar Car team joined the robotaxi challenge in Teknofest 2021 for the first time. At the beginning of development, they used a 2D LiDAR but switched to a 3D LiDAR to overcome the issues encountered because of outdoor use. They said,
We used a 2D LiDAR to make the barriers look like strips and to determine the appropriate steering angle to travel safely in between barriers. When we realized the 2D LiDAR was insufficient in outdoor conditions, we switched to a 3D LiDAR, but since our algorithms developed to use only with 2D LiDAR data, we used the data from 3D Lidar in 2D.
The team used their years of experience competing in solar car competitions and writing optimized codes to develop robust steering control algorithms. We’re looking forward to watching ITU’s Solar Car team in the future autonomous driving competitions.
The teams had a razor-sharp focus on what was necessary for completing the tasks in the contest, but they also tried new methods and techniques. All groups are eager to strengthen their teams further for next years’ more challenging competitions. All university teams have the same recruitment policy for every year’s competition team. Therefore, new members are recruited for every season, and former members serve as a consultant to their teams.
All teams we supported had a great ambition to develop and use algorithms to achieve a podium appearance. In the end, all teams we supported found a place on the podium.
Leo Drive will continue supporting up-and-coming self-driving car engineers to achieve their best in competitions and nurture their skills for real-world applications of self-driving. Soon, we will announce an academically affiliated program, where we will work with university lecturers to expose students to self-driving technology while undertaking their studies.
Consider following us on Medium and also on LinkedIn to stay posted.
