CL-EP-004: Panoramic Vision Perception Portable Experiment Kit

I. Experiment Box Overview

This experiment box is designed according to the core visual sensing technology of intelligent networked vehicles, focusing on the functionality demonstration of camera sensor modules. It includes a complete set of visual sensing experimental equipment, equipped with stereo cameras, capable of independently conducting camera experiments, teaching, and training.

II. Detailed Parameters of Vision Sensors

· Stereo Cameras:

o Sensor Model: IMX291, lens Size: 1/2.8 inches, suitable for high-definition image capture.

o Interface Type: USB3.0, ensuring fast data transfer.

o Effective Pixels: Up to 2 million pixels, resolution of 1920×1080, providing clear image quality.

o Image Format: Supports MJPEG/YUV2 (YUVY) format output, meeting various image processing needs.

o Frame Rate: Supports up to 50 frames/YUV/MJPEG at 1920×1080p, ensuring smooth video capture.

o Detection Targets: Capable of recognizing various target types such as vehicles, pedestrians, traffic signs, and traffic lights.

III. Display and Host Specifications

Display

o Size: 13.3 inches, providing sufficient visual display space.

o Resolution: 1920×1080, high-definition resolution delivers a delicate visual experience.

o Interface: Equipped with mini HDMI, 3.5mm headphone jack, built-in dual speakers, and HDR function, enhancing multimedia display effects.

o Power Requirement: DC12V, adaptable to various usage environments.

1. Host

o CPU: At least 6 cores and 12 threads, base frequency not less than 2.9GHz, and at least 12MB of third-level cache, meeting efficient data processing needs.

o GPU: Memory frequency not less than 1590MHz, memory capacity of at least 4GB DDR6, supporting high-load graphic processing tasks.

o Memory: No less than 8GB LPDDR4x 2666MHz, ensuring fast system response.

o Storage: Solid-state drive (SSD) capacity not less than 250GB, providing fast read/write speeds and ample storage space.

IV. Device Functionality

· Internal Parameter Calibration: The software supports camera internal parameter calibration, can generate calibration files, used to observe and correct camera distortion effects.

· Algorithm Support: The experiment box provides various vision recognition algorithms, such as YOLO target recognition algorithm, ROI lane-keeping algorithm, deep learning lane line recognition, and monocular distance measurement algorithm.

· Multifunctional Training: Supports loading different recognition algorithms, displaying different recognition functions through the interface, suitable for multi-project functional training.

· Data Input Diversity: Can process real-time camera data, recorded data packets, video images, and images output from decision-making and planning simulation benches, among other data sources.

In summary, this experiment box is a highly portable, comprehensive visual sensor teaching and training tool suitable for professional education and research in the field of intelligent networked vehicle technology. Through hands-on operation and experiments, learners can gain a deep understanding of the working principles of visual sensors in intelligent networked vehicles, data processing workflows, and their application in autonomous driving systems.