TfNSW M003 Demonstration

As per the project agreement, this Milestone is expected to involve:

Retrofitting of electric vehicle platforms with communication and unmanned driving capabilities.
Design and demonstration of interaction rules with VRU in pedestrian environments.

The task 1. can be considered to have already been completed.

Potential test locations:

Before testing can occur, the following must be conducted/verified:

Creating lanelet2 map with rules and structure allowing ease in satisfying the requirement/demonstration
Validating the localisation on the route that vehicle is expected to traverse

The outcome of the above can help inform the decision in:

developing navigation strategies suitable for the characteristics of the environment
whether placement of structures such as poles to aid localisation is required and their locations
finalising the demonstration outcome based on expected behaviour and environment characteristics

Currently, no data or map is available for the Crashlab site.

The following testing scenarios are designed to satisfy task 2:

Behaviour	Source of Object information
Vehicle slow down to a stop to wait for pedestrians at pedestrian crossing. Vehicle must resume to destination when the crossing is clear.	IRSU
Vehicle slow down to a stop to wait for pedestrians at pedestrian crossing. Vehicle must resume to destination when the crossing is clear.	Locally
Vehicle to navigate around pedestrians who are occupying and moving longitudinally along a lane if safe to do so.	IRSU
Vehicle to navigate around pedestrians who are occupying and moving longitudinally along a lane if safe to do so.	Locally

Note: fusion of object information from both IRSU and locally in the vehicle may be the subject of future milestone.

The functionality required to achieve the scenarios above must be generated and tracked on a package-level through issue board.

Create a map with localisation features and lanelet information
Incorporate pedestrian crossing as a special zone in lanelet2 map along with a speed limit
Policy for filling out objects detected on pedestrian crossing (if a person)
Detect road boundary based on laser data/camera image
Create obstacle polygon from currently observed road boundary
Create virtual lanelet based on locally observed road boundary/lane markings
Bias the local costmap by some offset based on round boundary
Incorporate speed information for different portion of paths based on underlying lanelet2 map in the planned path. Trajectory following controller to use this information to slow down in different areas.
Detector on the PX2 to extract bounding box of an object (person or car) for the tracking module
Next generation object tracker on the vehicle
Expand obstacle polygons of tracked objects (such as pedestrians or cars) to include future trajectory