Apex.Autonomy 1.3 released
We just released Apex.Autonomy 1.3 to our customers. Apex.Autonomy is a set of certifiable building blocks for developing autonomy functions. With this release, customers have a complete LIDAR based L4 autonomous driving stack, built from the ground up for safety-critical deployment.
Apex.Autonomy is offered separately from Apex.OS, the underlying ROS-based software framework. Together, Apex.OS and Apex.Autonomy offer the shortest path from prototype to deployment of a modern C++ codebase developed for real-time, deterministic performance.
This release includes a LIDAR-based NDT localization package to complement LIDAR perception. The localization module references a 3D point cloud map against real-world or simulated LIDAR data to determine the location of the vehicle in the world. A robust EKF-based method for velocity estimation and localization smoothens the results and additionally provides velocity estimates for planning and control modules.
Together, this release enables a vehicle to drive fully autonomously at low-speeds and stop for obstacles in its path. Routes are recorded by manually driving a path in a mapped environment like a parking lot or industrial zone. Routes can be followed autonomously using the building blocks included in this release.
Try out the open-source versions that we helped build in Autoware.Auto’s localization and prediction packages. Reach out to us at firstname.lastname@example.org if real-time, safety-certifiable, ROS-compatible autonomy software can accelerate your work!
The video below shows the Apex.Autonomy lidar-based localization and state estimation using a simulated car in a 3D-mapped space. Note the real-time, smooth performance of localization and velocity estimation, using off-the-shelf hardware including an x86 PC and Velodyne lidars.