JetsonHacks

Developing on NVIDIA® Jetson™ for AI on the Edge

Team ISF Löwen Simba 2 – Jetson based Autonomous Car in Carolo-Cup competition

Daniel Bräckelmann from Team ISF Löwen recently shared details of their vehicle used in the Carolo-Cup challenge earlier this month. Simba 2 won the hardest challenge of the competition, the Obstacle Circuit while placing third overall. Impressive to the Extreme. Looky here:

Background

The Carolo-Cup is an annual international event for university students held by “Technische Universität Braunschweig” in Brunswick, Germany. The competition is inspired by the DARPA Urban Challenge and features three different challenges:

  • Round Circuit – Drive as fast as possible on the circuit as possible for 3 minutes, while staying in the right lane of the course.
  • Round Circuit with Obstacles – Same as the first challenge, but obstacles are distributed over the course. Some of the obstacles are in motion.
  • Parallel Parking – Park as fast as possible in the smallest parking spot, parking in a large spot yields penalties.

In 2016, 15 student teams from all over Germany and a team from Gothenburg competed for two days in the Braunschweiger Stadthalle. The event is sponsored by many large German automotive manufacturers and part suppliers including Bosch, Infineon, Continental, BMW, VW and Daimler.

Simba 2

Löwen is German for “Lions”, which makes the name Simba, the lion cub from the film “Lion King”, a natural fit. The Simba 2 uses a stock Tamiya TRF417 for the chassis. Mounted to the chassis is a NVIDIA Jetson TK1 Development Kit, which is used for vision processing and algorithmic control of the car. Attached to the Jetson is a Teensy 3.2 µC for low level hardware interfacing with the steering servo, and the Electronic Speed Control (ESC) for the motor.

For the Carolo-Cup competition, a camera is mounted on a mast at the center of the car in addition to several other sensors (which appear to be IR and a laser line level). The camera is used for navigation, while the additional sensors are useful during the Parallel Parking part of the competition.

Team ISF Löwen has been working on the Simba project for around a year, with 10 students currently working on the project. You can sense the evolution of the project:

Simba Prototype and RACECAR
Simba Prototype and Simba 2 in competition trim

The software for the car was written from scratch using C++ and Qt. The resultant code is very lean (just about 5,000 lines of code).

A special ESC is used which allows very low speeds with high torque and fine-grain control. This is very useful in the Parallel Parking part of the competition:

Money quote:

For image processing we currently use OpenCV – CPU-only (Nvidias proprietary version – it’s just amazing how fast it is!).
Our plans for next years Carolo-Cup is to evaluate a complete GPU-based pipeline either using VisionWorks or the OpenCV GPU-module because we are close to the maximum performance the CPU can deliver (image size is 640×480 @ 50Hz).

Simba 2
Simba 2

There are also lights on the vehicle, which are used during the competition:

Simba 2 Lights
Simba 2 Lights

For more information on the Simba 2 build and to see Simba 2 in action, check out the Team ISF Löwen Facebook page and watch more of their videos on YouTube.

Simba 2 - Team ISF Löwen
Simba 2 – Team ISF Löwen
Facebook
Twitter
LinkedIn
Reddit
Email
Print

One Response

  1. Wow just catching up on jetsonhacks articles. These are much better than I thought they would be and I have 4 jetsons 🙂 A swarm going to post up to the local r/c board and see if I can pick up a few chassis cheap. Drone racing is getting large and this is sure to follow.

Leave a Reply

Your email address will not be published. Required fields are marked *

Disclaimer

Some links here are affiliate links. If you purchase through these links I will receive a small commission at no additional cost to you. As an Amazon Associate, I earn from qualifying purchases.

Books, Ideas & Other Curiosities