Fairchild was excited to participate in the Self Racing Cars’ first Autonomous Vehicle Track Day at Thunderhill Raceway a few hours north of our HQ in Willows, CA, two weeks ago. This new and unique new event was as much an engineering showdown as a race.
Our objectives were to test our motion tracking technology in high-performance automotive conditions and work with our Xsens colleagues to provide expertise on IMU technology to the other teams. Fairchild technology is incorporated in many of the participating teams’ vehicles, both autonomous and human-driven.
The event provided the opportunity for testing:
- Autonomous vehicles
- Drive train innovations (electric or otherwise)
- Sensors and cameras
- Software and algorithms
- Connected cars (v2v or v2i)
The event started with a group meeting that included a review of the three race groups – fully or partially Autonomous cars that got track time by themselves; Fast Humans featuring open passing for fast moving vehicles with trained drivers; and Slow Humans with passing only in straightaways and geared to cars and drivers without race experience.
As “Fast Humans,” we were assigned to the awesome Caryn Kealey, a driving instructor from the “Hooked on Driving” club, and her Mini.
My Xsens colleague Scott and I quickly began setting up Caryn’s car, starting with securing all the test equipment. It was no easy task strapping down five sensors, a USB hub, miles of cables, a data collection laptop, and a data logger Scott designed. The sensors (two MTi modules on Scott’s mount, an MTi, FIS1100 IMU and FMT1000 module on my mount, plus two GPS receivers) were Velcro-mounted to the floor mats and carpet.
Once everything was installed, we verified our programming and ensured the filter states were initialized for data acquisition.
When we finally accelerated onto the track, I hit a record and the sensors started logging data. Not only was this the first time our sensors have been around the track like this, it was also a first for me, what a rush!
We raced around the track, tires squealing, me pressed hard against the harness, and the sensors being pushed as hard as the car could. The course was very focused on turns, so the Mini, with its low center of gravity really excelled. We did 10 exhilarating laps in the 18 minutes allotted and I was hooked.
The data recorded was wonderful, with each turn written in accelerations and rates of turn. You could clearly see every time Caryn floored it and each time she slammed on the brakes at the last second to get the quickest time around the corner. From a data gathering perspective, you couldn’t ask for more!
Scott went on the second run and mounted his GoPro to the windshield. After his own wild ride, we checked the data and established our testing rhythm. We recorded each session and downloaded and analyzed the data while awaiting our next run.
It was quickly apparent that the high currents and a vehicle’s spinning metal render magnetic sensors useless. So we realized the first improvement we had to make to our sensor fusion filter was ditching the mag sensor! We also recognized that the 2m accuracy of current GPS technology is not enough for the autonomous driving application. Thankfully there are solutions in the 2 cm range and we hope to integrate this technology into upcoming products.
Most of the autonomous cars used our Xsens MTi modules and the teams loved the performance. In the future though, they all expressed the need for something more like the small FMT modules with their attractive entry price. Everyone was very interested in the accuracy improvements in FMT development and look forward to benchmarking with the MTi 700 series.
Scott and I met more autonomous racing enthusiasts throughout the day and they gave us their wish-list of future needs, which comes down to three things – low cost, high performance, and automotive rating. Meeting these requirements is really what is needed for a MEMS + Sensor Fusion solution to help the industry transition from small companies testing kits too expensive for production volume to having every car equipped with an autonomous button…or seeing cars race themselves around a track for our entertainment.
We left the track with hours of test data, pounds of brake dust, and excitement for what will come in the next few years in the racing/driving space.
Our IMU product specialist Scott was able to support and observe track tests with some of these solutions:
- Autonomous Stuff’s autonomous Lincoln MKZ using an Xsens MTi-G-710
- Denso’s autonomous Tesla Model S using an Xsens MTi-G-700
- PolySync’s Kia measuring with an Xsens MTi-G-700
- In addition, Xsens completed extensive testing with an MTi-G-710 and MTi-3 in a Mini Cooper
Self Racing Cars, autonomous vehicles, and related technologies are markets where Xsens technology has a firm foothold with high-performance 3D motion tracking solutions. Fairchild’s mass market motion tracking devices are making inroads with the autonomous innovators and events like the Autonomous Vehicle Track Day will help us push our technology even further and help companies operating in this field develop innovative and safer systems. I can’t wait to get back out on the track for more testing!