Your behavior appears to be a little unusual. Please verify that you are not a bot.


Software steers autonomous vehicle testing

November 21, 2017  - By

Assessing the performance of autonomous systems under real-world conditions requires an ultra-precise ground truth reference against which to benchmark vehicle performance. A GNSS-plus-inertial post-processing software can provide this capability, taking real-time GNSS data — which are subject to outages, obstructions, weather-induced errors and more — from the vehicle and correcting the solution. This can improve meter-level data to centimeter-level, a critical standard for safe autonomous performance. A free webinar on Nov. 30 gives both a high-level overview and close-in details of this process.

Autonomous vehicle testing requires ultra-precise ground truth.

Many sub-systems must function flawlessly and interact seamlessly for safe autonomous vehicle performance.  Fielding such a vehicle requires rigorous testing, repeated many times; this in turn requires close comparison of the vehicle’s real-time GNSS data to a ground truth of its performance. Post-processing software that combines GNSS with inertial navigation system (INS) data, to bridge GNSS outages common in real-world driving, can provide this capability. Whether the tests are evaluating potential sensor suites, benchmarking their own solutions, or generating high-definition maps, post processing maximizes the accuracy of the solution by processing previously stored GNSS and INS data forward and reverse in time, and combining the results.

Novatel’s Waypoint software package, Inertial Explorer, offers this capability, whether lower-grade or high-end inertial sensors are employed. An examination of the process is afforded in the free webinar, from the converging viewpoints of three speakers:

Steven Waslander, associate professor at the University of Waterloo, heads a project collecting 1,000 km of data in all-weather conditions for a new public road driving dataset focused on autonomous driving challenges. He directs the Waterloo Autonomous Vehicle Laboratory (WAVELab), extending the state of the art in autonomous drones and autonomous driving through advances in localization and mapping, object detection and tracking, integrated planning and control methods and multi-robot coordination.

Terry Lamprecht, director of products at AutonomouStuff, a supplier of components, services and software that enable autonomy, will discuss verifying proper installation, and creating a baseline data set to benchmark against data collected on autonomous vehicles in real-time.

Natasha Wong Ken, product manager at Waypoint, will give a high-level technical overview of post-processing techniques and settings, including forward and reverse processing, tightly vs. loosely coupled, PPP vs. differential, and more.

Registration for the November 30 webinar is free. For those not able to attend the live broadcast, all audio and presentation slide components can be downloaded after air date for viewing at convenience.

Some of the new capabilities explored jointly by NovAtel and AutonomouStuff are covered in the August cover story, Autonomous Assembled.