v1.4.0

✨ Add Guardian node
✨ Add working CARLA simulation agent (carla.launch.py)
✨ Add junction management node (basic FSM)
✨ Add grid summation node
✨ Add RTP, the recursive tree planner, with inbuilt motion control
✨ Add airbags, a low-level emergency braking system

Numerous other refinements and integrations in preparation for real-world use!

v1.3.0

Navigator can now be developed and run entirely in Docker.

This is essentially a refined version of Navigator's previous release. We cleaned up many of our old/faulty packages that weren't in use. With Docker successfully implemented, development is more streamlined.

This version adds:

• Docker (ROS/Navigator/CARLA Leaderboard materials & dependencies included), instructions found in our documentation below
• ROS RMW set to CycloneDDS from FastRTPS, finally allowing for docker containers
• New documentation, including instructions to run docker.

This version removes:

• Old, stale packages

v1.2.0

This is a stable version of Navigator "Classic," which was purpose-built for Demo 2. The algorithms here are purely deterministic/classical, and feature no ML.

This version adds:

• Major improvements to our behavior planner, motion planner, and zone handlers
• A custom sim_bridge for interfacing with CARLA
• New visuals for calculated paths, zones, and trajectories
• A small library and integrations for the OpenDRIVE HD map format
• New documentation that lives directly in our repo

This version removes:

• Support for the LGSVL simulator (we've transitioned to CARLA)
• Support for the Lanelet HD map format
• Old, stale packages from e.g. Autoware.Auto (we're grateful for Autoware's support!)

v1.1.0

This version includes no new functionality, but lays the groundwork for great things to come. Changes in this version include automatic testing, an actual README, some information on our custom hardware design, a GPS node, and a basic safety manager.

Demo 1

At this point, the car is able to steer itself through certain environments on a predefined path. This is our first actual autonomous drive, and lays the groundwork for the rest of our project.