Project In-cabin: A New Vision for Driver Monitoring Systems

In the rapidly evolving world of autonomous technology, the focus is often on how a vehicle interacts with its external environment. We hear about sensors that detect pedestrians, map roads with lasers, and navigate through complex traffic. However, a critical part of this technological leap happens on the inside: the cabin. This is the domain of the Project In-cabin, an initiative by Eye2Drive to revolutionize how we monitor drivers and ensure safety.

Eye2Drive, a company that designs CMOS image sensors inspired by the human eye, is the force behind this project. The company’s unique approach to sensor technology mimics the human eye’s ability to rapidly and dynamically adapt to vast changes in light conditions. This is crucial for autonomous systems that need to perceive their environment accurately, and it’s particularly vital for a system focused on the driver.

The Core Objectives

For Project In-cabin, the primary goal is to configure the sensor to achieve the highest possible graphical image quality of the driver. This is similar to Eye2Drive’s other project, “Controsole” (Against the Sun), which focuses on external environments such as the sea and the sky. The objective is clear and twofold:

1. Always Visible: The driver must be visible at all times, regardless of the lighting conditions inside and outside the vehicle.
2. High-Quality Imaging: The image quality must be good enough to support downstream tasks, such as object recognition.

To achieve this, the project involves a significant phase of defining and building “voltage ramps”. These ramps are essentially sensor configurations that dictate how the sensor behaves under different conditions. For instance, a voltage ramp might be optimized for normal light conditions, while another might be better suited for extreme lighting, such as when the sun is absent or “blasting”.

The ultimate goal is to create a sensor that can automatically adapt to these varying conditions. This is managed by a sophisticated auto-regulation algorithm, which is currently 100% AI-driven and needs to be developed and adapted for this specific scenario.

The Project In-cabin Challenge

The automotive cabin poses unique challenges for a vision system. A sensor pointing at the driver must contend with a variety of difficult and rapidly changing lighting conditions. These include:

• Tunnel Transitions: The sudden shift from bright sunlight to tunnel darkness can momentarily blind a conventional camera, leaving the driver invisible to monitoring systems. At Eye2Drive, we design technology to perform real-time adaptive HDR (High Dynamic Range) and sensitivity, ensuring a seamless transition and continuous visibility.
• Sun Glare: Sunlight entering through the rear window or side window can create intense glare, obscuring the driver’s face. Eye2Drive’s sensors are designed to acquire images natively with the desired dynamic range characteristics. They are effectively mitigating this glare without resorting to heavy post-processing that can introduce artifacts such as ghosting.
• Flickering Lights: Flickering can be caused by rapid cycling of LED lights, such as those in traffic signals or digital dashboards. These can confuse conventional sensors, leading to missing information. Eye2Drive’s technology is intrinsically immune to this problem, ensuring all information is captured accurately.

These challenges highlight the limitations of traditional camera sensors, which often suffer from over- and underexposure, ghosting, and flickering. Eye2Drive’s bio-inspired approach is a direct solution to these problems, providing a more robust and reliable foundation for in-cabin monitoring.

The Benefits of a Flawless “Eye”

The success of the Project In-cabin directly enables the functionality of Driver Monitoring Systems (DMS). These systems are crucial for safety in both human-driven and semi-autonomous vehicles. The goal is to ensure the driver is always visible and the image quality is high. The sensor allows these systems to perform a variety of safety-critical tasks, such as:

• Detecting Driver Inattention: DMS can monitor the driver’s head and eye movements to detect signs of distraction or drowsiness. If the driver is looking away from the road or their eyelids are drooping, the system can provide an alert.
• Assessing Stress and Fatigue: By analyzing facial expressions and other subtle cues, DMS can gauge the driver’s stress levels. This can be a vital input for systems designed to prevent accidents caused by driver fatigue or emotional distress.
• Enabling Higher Levels of Autonomy: As vehicles move towards Level 3 and Level 4 autonomy, the DMS becomes even more important. But while the vehicle may handle most driving tasks, the driver must be ready to take over when requested. A reliable DMS ensures the driver is alert and ready to regain control at a moment’s notice.

Without a sensor that can consistently and reliably capture the driver. Without a sensor, these systems are useless. Project In-cabin addresses this fundamental “blind spot” in the algorithm, making DMS and other advanced safety features truly viable.

Beyond the Cabin: A Broader Vision

The technology developed for Project In-cabin is part of Eye2Drive’s broader vision to empower the next generation of intelligent vision systems. The company’s sensors are not just for cars. We design them for a wide range of autonomous navigation systems. Our sensors can be used for drones, industrial robots, and submarines.

The core of their innovation lies in their AI-ready hardware solution. This means we design the sensor for direct interaction with AI and machine learning algorithms. This can dynamically shape the sensor’s output by tuning simple input parameters. This design philosophy simplifies the perception pipeline, reduces the computational burden, and enables more reliable feature extraction.

In a world where current solutions often rely on heavy software processing and hardware redundancy to compensate for sensor limitations, Eye2Drive’s approach offers a more elegant and efficient path forward. Their technology’s ability to natively handle challenging conditions and mitigate common artifacts positions them as a key player in the deep-tech ecosystem. The Project In-cabin project is a perfect example of this, demonstrating how bio-inspired technology can solve a critical, real-world problem and pave the way for a safer, more autonomous future.