XR-F3 ROBOT Raspberry Pi autonomous driving car

Intelligent transportation   Deep learning   Neural network   Visual recognition

Core Highlights

XR-F3 Raspberry Pi Autonomous Driving Car is an autonomous driving car that combines deep learning, neural networks, and machine vision technologies. Adopting a front wheel steering mechanism similar to that of automobiles, namely the Ackermann chassis design, ensures its driving stability and efficient steering performance, making its autonomous driving mechanism similar to current popular new energy vehicles. It has become one of the popular autonomous driving projects in academia, technology enterprises, and research institutions.

Not only is it suitable for students majoring in electronic information, automation, and artificial intelligence to learn neural network algorithms, deep learning techniques, and Linux development techniques, but it also aims to deepen users' understanding of artificial intelligence through hands-on practice and engaging experiences.

This tutorial starts with understanding the hardware of small cars, covering basic usage methods, introductory secondary development, exploring the implementation of combining neural networks with hardware, and gradually guiding you into the world of development and deep learning technology.

Course List

RASPBERRY PI with built-in Raspberry Pi 4B

AI performance significantly improves, image to video transmission becomes smoother

CPU 64 bit 1.5GHz quad core (28nm process)                 GPU   Broadcom VideoCore  VI@500MHZ

Ackermann steering structure

Ackermann steering mechanism, as the cornerstone of modern automotive design, is a widely adopted standard technology in the new energy vehicle market, particularly favored in the field of autonomous driving. This structure achieves precise steering by adjusting the angle difference between the inner and outer wheels, known as the Ackermann steering principle. The XR-F3 Raspberry Pi autonomous driving car introduces this concept and achieves a 100% Ackermann rate, which means that when turning, the angle of the inner wheel will be greater than that of the outer wheel. This rigorous design not only improves the accuracy of steering, but also greatly promotes the efficiency of robot learning and research.

Model training for autonomous driving

The XR-F3 Raspberry Pi autonomous driving car is equipped with the TensorFlow neural network framework, which seems to endow it with thinking abilities similar to those of the human brain. During its operation, the car captures real-time video footage through a camera and records the speed (i.e. throttle amount) and steering angle corresponding to each frame. After accumulating sufficient data, this raw information will be input into the Keras artificial intelligence neural network library, and then processed through the TensorFlow framework to train accurate model files.

                            Original video                Retrieve video frames through OpenCV        Processing video frames to detect objects

1. Sand table autonomous driving

In autonomous driving mode, the robot uses real-time footage captured by its camera to compare with the data weights stored in the model file, and based on this, determines the precise amount of throttle and the specific angle of steering.

2. Smart traffic sign recognition

Through the training of deep learning model base, the auto drive system can make accurate comprehensive decisions on road conditions according to such factors as lanes, traffic signs and traffic lights.

                                       Identification of humanoid crosswalks                                                            Stop sign recognition

                                             Traffic light recognition                                                                             OLED parameter display     

Intelligent collision prevention

In autonomous driving mode, the ultrasonic collision avoidance function can ensure the safe driving of the car. The ultrasonic sensor assembled at the front end continuously scans for obstacles ahead and automatically stops when it is below a safe distance. This function effectively prevents rear end collisions or collisions with fences caused by deviation from the track when multiple vehicles are driving, protecting the vehicle body and surrounding environment.

Cross platform interconnection control

Optional autonomous driving sand table

Optional autonomous driving sand table

              Autonomous driving map sand table              Traffic light model                        Street lamp labeling * 6

                                                                                           (Can manually control the light change)

Application scenarios

               Teaching and training                                Competitive competition                                Autonomous driving

Hardware distribution

PWR.A53 Robot Drive Expansion Board

Product List

                   Standard version: ◎                                                                                      Autonomous driving kit:◎

                                   Raspberry Pi autonomous driving car                                         Autonomous driving sand table

                                            Charger                                        Traffic light model                              Street lamp labeling * 6