CamAlien

CamAlien is a high-speed camera system designed for efficient monitoring of vegetation and objects along roadways. The camera system captures high-quality images at rapid intervals, to monitor the full roadside when following traffic speed. The system utilizes a global shutter sensor for accurate scene rendering and geotagging for precise data collection. The camera system makes it possible to collect extensive data sets, which can be used for a digital library of road object types and locations, but also subsequent training of machine learning algorithms for automated object recognition.

CamAlien consists of a camera mounted on an adjustable arm, a real-time GPU-accelerated processing unit, remote controller, and a car mount. Furthermore, it has two types of power connectors, allowing it to be powered either from a car with 12 VDC or from a socket with 100–240 VAC. The physical remote controller allows easy control and simple monitoring of the system.

The processing unit of the system includes a real-time GPU-accelerated compute element that is connected to the camera, the GNSS module, the modem, and the remote controller. The figure below illustrates with a flowchart how different components of the system are connected, and how images flow from the camera, through the memory (RAM) and an internal queue, and are finally stored at an external SSD.

A buffer allows the user to record images for a configured number of seconds back in time, when a button is pressed in either the web interface or with the physical remote controller. When recording is active, raw images are stored on a fast internal NVME disk. From here, they are anonymized asynchronously using a pre-installed AI model that automatically detects and removes people, bikes, and cars from the images. As the system includes a real-time GPU-accelerated Nvidia Jetson AGX Orin compute unit, it is furthermore possible to replace the pre-installed AI model with custom models running in real-time. Finally, the anonymized images are stored to an external SSD.

The camera system is designed for high speed, up to 130 km/h. To get high-quality images at these speeds, a global shutter camera is needed, and the exposure time needs to be short to ensure a low amount of pixel blur. The graph illustrates the relation between speed, exposure time and perceived pixel blur.

CamAlien automatically adjusts the exposure time based on the current speed to obtain the best possible image quality. Furthermore, the frame rate is adjusted accordingly, to obtain full coverage of the road side at a configurable distance. That is, the system is designed to maintain a small overlap between consecutive images.