Autonomous Emergency Navigation to a Safe Roadside Location

In this project, we developed essential modules for achieving the proposed autonomous emergency navigation function for an automated vehicle. We investigated and designed sensing solutions for safe roadside location identification, as well as control solutions for autonomous navigation to the identified location. Sensing capabilities are achieved by advanced fusion algorithms of 3D Lidar and stereo camera data. A novel control design, based on dynamic differential programming, was developed to efficiently plan navigation trajectories while dealing with computation delay and modelling errors. Preliminary validation of proposed solutions was carried out in a simulated environment. The results show strong potential for success, especially for the control module. Hardware integration in a real vehicle has been ongoing in a parallel fashion to enable field tests of developed modules in future work. Key sensing equipment was installed and calibrated and used to collect data for offline analysis. The retrofitting of the vehicle’s actuation mechanism was finished with the whole drive-by-wire system in place. Future work will involve road testing the developed systems.