Through the mechanical decoupling of the input device, the steering actuators and the drivetrain (X-by-wire) various designs of the human machine interface are possible concerning the generation of commanded signals on the one hand and the choice of signals displayed on the haptic input device via force feedback on the other hand. Haptic feedback improves the perception of the dynamic properties of teleoperated mobile systems and the external forces acting on those robots. The various sensors of the ROboMObil are able to measure e.g. wheel forces, vehicle acceleration or the vehicle slip angle. Thus a huge variety of relevant signals can be displayed as a feedback on the sidestick to improve the vehicle perception compared to the forces resulting out of the mechanical coupling in a conventional vehicle.
In order to validate newly developed driving stability control approaches safely, the ROMO shall be teleoperated during those test drives. Since an operator in a teleoperation station does not obtain information on the vehicle's dynamic state through his vestibular system in contrast to a operator in the cockpit, the relevant haptic feedback design differs between the two operator locations. Time delay and package loss in the communication channel of the teleoperation system can have a destabilizing effect on the control circuit. The energy-based “Time Domain Passivity Approach” is applied to solve this problem rendering a complicated modelling of the hardware unnecessary.
A fully autonomous car is still a future vision and even then a human driver wants to drive manually from time to time. An intermediate level between full autonomy and a manual operation mode is the so-called shared autonomy, which has been successfully demonstrated in space robotics. In case of the ROMO shared autonomy means that a human operator is making decisions on the highest level of vehicle navigation whereas the artificial intelligence overtakes track keeping and stabilization. The main difference to mobile robotics is that the operator does not interact with the artificial intelligence under static conditions like e.g. during the planning of trajectories but online under dynamic conditions.