Sample prediction module for a vehicle in the simulation framework.

Either forwards the perceived objects without predicting their motion (no_prediction.launch) or predicts vehicles with constant velocity along the centerline of possible lanelet sequences and other objects with constant position (lane_prediction.launch).

• this package is part of the simulation framework
• see coincarsim_getting_started for installation and more details

• started within the a vehicle launchfile of the simulation_initialization_ros_tool

• parameters that need to be passed to the launchfile sample_prediction.launch:

  • vehicle_id: Id of the vehicle, needs to be unique within the framework

  • vehicle_ns: Namespace of the vehicle, needs to be unique within the framework

  • perc_pred_obj_topic: Topic for the perceived objects

  • pred_plan_obj_topic: Topic for the predicted objects

• the lanelet2_map (for lane_prediction.launch) is retrieved via the package lanelet2_interface_ros

• fork this repo
• use your own algorithms for predicting the motion of perceived objects
• ensure that
  • /$(arg vehicle_ns)/$(arg pred_plan_obj_topic) is published
  • all internal ROS communication stays within the prediction namespace

Pascal Böhmler, Nick Engelhardt, Tobias Kronauer, Alexander Naumann, Maximilian Naumann

This package is distributed under the 3-Clause BSD License, see LICENSE.