In everyday traffic, human road users communicate intuitively by small gestures: a short wave or nod is often enough to indicate that the other person is allowed to pass through. This element of communication is missing in autonomous cars. Ford is now testing how such a simple "gesture language" between robocar and human road users could look like.
When Daimler sent an autonomous Mercedes through the traffic in 2013, a fundamental weakness of the electronic chauffeurs became obvious: An elderly couple was waiting at a pedestrian crossing. The automatic Mercedes stopped allow the pedestrian to cross. However, the two pedestrians made a gesture to make it clear that they would prefer the car to drive first. The computer on board the Mercedes didn't understand the hint and insisted on not continuing until after the pedestrians. The deadlock situation could only be resolved after the human safety driver intervened.
This problem is still waiting for a solution. Wherever in the world autonomous cars are particiupting in normal traffic, such misunderstandings occur. Now Ford is working with the Virginia Tech Transport Transportation Institute to establish an interaction between human and electronic road users by providing easy-to-understand signals. "Understanding how self-driven vehicles can operate in the real world is the basis for the development of future traffic reality," says John Shutko, Human Factors Technical Specialist, Ford Motor Company. "We must find solutions to the challenge of ensuring that at some point no human driver will be behind the wheel. It is about how we can replace natural human gestures to ensure the safe and efficient operation of self-propelled vehicles in public spaces".
Initially, the researchers considered displaying large textual information on the vehicle, but this would require that all road users understand the same language. The use of symbols was rejected because new symbols are not recognized enough by humans. In the end, they experimented with light signals. Light signals for turning and brake indication are already standard and generally accepted, so that a lighting application is considered the most effective means of communication.
For example, a self-driving vehicle could signal whether it is operating in autonomous driving mode, whether it wants to start or whether it wants to remain stationary. Towards this end, Ford positioned a light bar on the windshield of a test vehicle. In addition, six HD cameras were installed to detect the behavior of other road users in a 360-degree panoramic view.