Teleoperation: Remote Assistance Using Collaborative Trajectory Planning

Summary

The approach of distributed behavioral decision-making between an automated vehicle and a human remote operator is made tangible. The remote operator intervenes to provide support if the vehicle is unable to find an independent solution. In the developed system, humans and automation work together by the operator adapting the planner’s boundary conditions. Two central use cases are demonstrated: the release of a non-classifiable traffic light and the extension of the drivable area when the ego lane is blocked.

State of the Art

Existing publications on teleoperation mainly focus on the direct control of automated vehicles through continuous input. Remote assistance, on the other hand, makes it possible to optimize the teleoperation process through targeted, discrete inputs and to better exploit the potential of automated vehicles. Targeted interaction with individual modules such as planning or perception can significantly improve the safety and efficiency of teleoperation. So far, however, remote assistance for automated vehicles has only been covered in a few publications – mostly as part of simulation-based studies.

Technological Innovations

For the first time, a teleoperation solution will be presented in which a remote operator supports an automated vehicle from a distance by specifically adapting the planner’s boundary conditions – without any direct control via steering wheel or pedals. The remote assistance will be demonstrated using an automated vehicle, whereby the connection to the teleoperation system is made via arbitration graphs, which enable a seamless connection between the two subsystems.