(unmanned one) software architecture _ software for unmanned driving systems

I. Overall structure

Based on my understanding of unmanned driving system, the software architecture of unmanned driving system is layered, which is roughly divided into 3 layers: perceptual layer, decision-making level and control layer. Second, the perceptual layer

The perceptual layer is mainly realized by various sensors and high-precision maps, including the location of the vehicle and the recognition of the object.
Vehicle positioning is mainly through the optical radar (LIDAR), GPS, inertial sensors, high-precision maps and so on the synthesis of information to obtain the exact location of the vehicle, its positioning accuracy can reach the CM level;
The object recognition mainly uses the light radar and the binocular camera realization; III. decision-making

The input of the decision-making level includes the information of the perceptual layer, the planning of the path and the feedback from the control layer, and the decision instruction is issued by increasing the learning algorithm. Decision instructions include: Following the car, overtaking, speeding, braking, deceleration, steering, turning, etc. four, control layer

According to the orders issued by the decision-making level, the control layer on the vehicle implementation of specific control, including: throttle control, brake control, steering wheel control and the control of the gear;