Continental’s auto-driving software gets handover issue under control

July 17, 2017 // By Christoph Hammerschmidt
One of the most difficult challenges in autonomous driving is mastering the vehicle’s behavior in the time span between a possible failure and the takeover by the human driver. Car electronics supplier Continental now claims to have a solution for this problem.

Continental has developed an algorithm it calls Cruising Chauffeur. Running on the vendor’s Assisted and Automated Driving Control Unit (ADCU), the Cruising Chauffeur processes the data from the vehicle’s environmental sensors such as camera, radar and lidar. The software transforms these data into a 360° environmental model of the vehicle. This model is merged with the data from a high-resolution real-time road map that contains all moving and static objects as well as the course of the road and the traffic lanes. In this model, the vehicle’s position is determined and updated continuously. Thus, the algorithms can identify areas that can be safely used by the car within the boundaries of the traffic rules and the driving task and adjust the trajectory accordingly. This enables the Cruising Chauffeur to change lanes automatically; it also can pass another car and after completion return to the standard lane. These algorithms are designed for automated cruising on highways – a relatively simple task since typically the automatic driver does not have to handle oncoming traffic or pedestrians. All in all, this is what describes level 3 of autonomous driving.

But as soon as the vehicle is approaching If the end of the motorway section becomes closer, the vehicle has to request the human driver to take over. As long as the driver is closely monitoring the driving action in accordance with the requirements of level 3 automatedd riving, this is not a problem; being in a standby position, the driver can take over within a few seconds. The transition to level 4 of automated driving however is not as easy: In L4, drivers are allowed to devote themselves to other activities and even take a nap. Thus, they get out of the context and cannot drop in within a few seconds.

To handle this issue, Continental is using an interior camera that monitors the driver’s position and analyzes his direction of view. Smart algorithms detect the driver’s alertness and its capability to assume the task of driving. If the vehicle is approaching further the point where it is intended to leave the motorway and the driver is still not ready, the optical and acoustic warning signals are intensified. If the driver still does not show any reaction, the Cruising Chauffeur initiates the Minimum Risk Maneuver – which means that the computer either enters the emergency lane and brings the vehicle to a safe stop, or, if no emergency lane exists, it activates the hazard lights and stops where possible; if necessary it continues the ride at low (and decelerating) speed until it finds a suitable spot to stop.