Localization for the Next Generation of Autonomous Vehicles

  • Room: C142
Wednesday, May 10, 2017: 10:30 AM - 11:00 AM


Chris Osterwood
Carnegie Robotics
Fergus Noble
Swift Navigation


The next generation of autonomous systems, from commercial UAVs to outdoor ground robots to self-driving cars, will need much better localization accuracy than has traditionally been available. Additionally, high-volume autonomous systems and vehicles require a solution that is available at an acceptable price point. There has also been a lack of consensus around what combination of sensors will be required, especially for safety-critical applications that have to work in a wide range of conditions. In this session, we discuss the tradeoffs of varying sensor packages and demonstrate that precise GNSS, coupled with an IMU, provides the localization cornerstone for these systems. Precise knowledge of system position and attitude increases the utility of locally collected imagery and 3D sensors by reducing the computational cost of registering data with existing maps or previously collected data. Precise GNSS provides the system with global position while the IMU provides heading, pitch, and roll information while also allowing global position to be known by the system at 100 Hz or more. We will present results from a low-cost Inertial Navigation System (INS) that incorporates RTK GNSS and a MEMS IMU, showing that it provides excellent performance under a range of challenging conditions and is available at a fleet-friendly price-point.


Who Should Attend

In general, the target audience for this session is engineering leaders (team leads to executives) from companies that are designing—or interested in designing—autonomous vehicles of all kinds such as UAVs, autonomous cars, ground robotics, etc. This talk is applicable to any vertical from agriculture to automotive.

Specifically we are targeting people who are responsible for choosing and integrating sensors which provide navigation for these systems. The talk will focus on GNSS and inertial navigation and therefore the talk will be of particular interest to those who are considering, and currently, using these technologies.