In recent years, significant research has been directed at solving the localization problem of autonomous driving. The navigation system required of a fully autonomous platform must provide high-rate accurate positioning in all environments. To achieve these requirements and enable widespread adoption, multiple sensors must be used within an affordable range.
INS, and GNSS are two powerful positioning technologies that on their own, can provide super accurate positioning data for the user. However, pure INS systems are typically used for short durations as they accumulate errors over time, and GNSS signals to weaken or drop all together in environments like tunnels or in dense urban centers.
Typically, INS and GNSS are integrated to achieve lane level accuracy for autonomous vehicles in open sky environments. However, to achieve continuous and accurate positioning in even the most challenging environments, perception sensors like cameras, lidar, or radar may be utilized as another source of positioning information.
The integration of all three of these technologies allows one signal to fill in the gaps when another signal may be lost or disrupted. This tight integration provides the system with a contingency, resulting in continuous positioning data in any environment.
This presentation provides an overview of different positioning approaches, followed by why INS is the core for navigation with the AUTO software. Building on this, we introduce the AUTO software utilizing the tight integration of INS/GNSS paired with Radar.
To demonstrate, a test setup using a multi-radar configuration is shown. The initial results focus on multi-radar results for vehicle. Following those results, we look at the results for autonomous robotic platforms.