Isaac is an end-to-end (E2E) autonomy stack aimed at enabling a fleet of coordinated robots that function robustly and safely in large, highly-dynamic, unstructured environments among humans.

Isaac version 2.0 is the first step in this direction.

Isaac 2.0 Early Access features a lidar-based autonomous navigation stack that contains a Global Route Planner, Mission Client, Behavior Planner, Lidar Localization (global and continuous), and Wheel-IMU Odometry, as well as a new Path and Trajectory Planner and Controller. This release also includes tools for Data Collection and a cloud-based Map Creation Service.


Figure: Isaac 2.0 EA Software Stack

This release enables autonomous navigation for a robot to navigate large open spaces, narrow corridors, and narrow/cluttered passages with unmapped obstacles.

The Isaac 2.0 Early Access release allows you to perform these actions:

  1. Create a Map using Carter Robot Version 2.3 (a differential wheel

    reference robot with a subset of NOVA Sensors and Jetson AGX Orin computer)

    1. Use Carter v2.3 to collect data for a large environment.

    2. Upload the data and obtain maps (Voxel Map and Occupancy Map) in a


  2. Use the map and Navigation Stack on Carter for the targeted ODDs

    1. Use the Carter v2.3, map and Isaac Navigation Stack to trial lidar

      perception, semantic awareness (maps), and the advanced path planner.

© Copyright 2018-2023, NVIDIA Corporation. Last updated on Oct 30, 2023.