top of page

Coursa Drive: Integrated Navigation Solution for Autonomous Vehicles




CASE STUDY


Accurate, reliable, and continuous positioning is essential for Autonomous vehicle navigation applications. Precise positioning algorithms based on Real-Time Kinematics (RTK) using Global Navigation Satellite System (GNSS) are widely used for this purpose and can provide a centimeter level accuracy in open sky environments. Positioning availability and accuracy are subject to degradation due to the different environmental aspects and GNSS satellite availability. To help alleviate these problems, Coursa Drive, a multi-sensor integration system, is proposed to provide a seamless navigation solution.


Coursa Drive integrates an RTKGNSS with an Inertial Navigation System (INS) to correct for sensor errors and calibrate the INS when an accurate RTK GNSS solution is available. Different sensors are added to the integrated solution to help reducing the error growth during outage intervals such as odometer and barometer. The integrated solution provides continuous, reliable, and accurate solution consisting of position, velocity, attitude and heading together with all their accuracy measures to the autonomous vehicle system. Coursa Drive can provide decimeter position precision for short periods when the GNSS or perception systems are uncertain or unavailable.


Coursa Drive is intensively tested in different environments, with widely different weather and temperature characteristics to assure robustness and reliability across various conditions. The testing included a variety of simulated and natural GNSS outages to provide statically significant (for each outage duration) key performance indices (KPI) in those outages. For high-speed scenarios on highways needed for level 2 and level 3 of autonomous vehicles, and during full blockage of GNSS signals in both natural or simulated outages, Course Drive enhances the sensor-only navigation solution and vehicle positioning to <0.2% error of the travelled distance.


Read the full paper here.

Comments


bottom of page