MILD System Enhances Human-Vehicle Collaboration with Bidirectional Perception
A new research paper introduces the Mediator-in-the-Loop-Driving (MILD) system, designed to address cognitive demands in partial driving automation by improving bidirectional understanding between human drivers and vehicles. The system uses an agentic architecture with a perception agent for joint in-cabin and out-of-cabin awareness and a lightweight strategy agent for compliant, explainable actions. The paper argues for shifting the human role from passive supervisor to active manager.
Key facts
- Partial driving automation can increase cognitive demands on human drivers.
- Lack of transparent insight into vehicle intentions and limited awareness of driver state cause misalignment.
- MILD system is based on an agentic system architecture.
- MILD integrates a perception agent and a lightweight strategy agent.
- The system aims for synergistic human-vehicle collaboration.
- The paper proposes a paradigm shift to elevate humans to active managers.
Entities
Institutions
- arXiv