As the industry pushes toward more advanced driver assistance and higher levels of automation, the challenge is no longer just technical performance, it’s how humans and automation work together.
In this technical deep dive, Robert Fuchs explores the concept of collaborative steering, examining how haptic shared control and admittance control can strengthen driver–automation teaming while preserving performance, safety, and driver engagement.
Written by
Robert Fuchs
Head of R&D Department
How does collaborative steering differ from traditional steering-based ADAS in terms of control authority and driver intent integration?
Two types of driver interaction with the steering are available in partial automation. First, the steering is used as a hard switch. The driver cannot intervene but regains manual control if applying a torque sufficiently high to the steering wheel. Manual and automated driving are separated increasing the risk of driver disengagement.
Second is a more interactive steering. Typically, this is achieved by reducing the steering angle controller gains. Interactivity is gained at the expense of reduced tracking performance. Driver engagement is promoted.
The preformance obtained is acceptable for straight line driving conditions but when the ODD expands to all roads conditions accurate tracking and interactivity are both required.
Collaborative steering enables driver intervention and provides meaningful haptic feedback without deactivation of the automation and with no limitation of the ODD.
What role do haptic shared control and admittance control play in improving driver–automation teaming?
The type of steering control gives different interpretations haptic shared control. Available interactive steering control literally share the control. Driver and automation have their share of the steering task. Efforts from both are required to achieve consistent performance.
Admittance control provides a different interpretation of haptic shared control in the sense that the automation is always active and able to track accurately any planned trajectory. The driver is an independent agent that can intervene any time to adjust the vehicle position.
How do you balance objective ADAS performance metrics with subjective human factors like workload and frustration?
This is the essence of human-centered design where ADAS and driver performance should be optimized jointly. Thanks to admittance control, the tuning of vehicle tracking performance and haptic are independent. Tracking performance is always maximized while haptic tuning should accuratly reflect system performance and risk without becoming too intrusive.
What does “human-centred” design really mean in the context of shared vehicle control and where do systems most often get it wrong today?
As indicated, human-centered design is about the joint optimization of ADAS and driver performance. There are few ADAS that are not robust to misuse or disuse. For example, automated lane change is triggered upon activation of the turn signal and verification of the driver hands on the steering wheel, which do not guarantee that the driver has checked for no incoming traffic.
Automated parking is another function that is available for already 15 years but that very few drivers use. This is likely an example of disuse caused by anxiety.
Why are forums like AutoSens/InCabin so important for the industry?
It is a necessary platform to communicate and share new knowledge and understanding necessary to shape the future of automotive and traffic safety.
Want More Interviews? Watch them on InCabin ⬇
