The invention relates to an automatic parking method and system based on generative adversarial imitation learning, and the method comprises the steps: generating a corresponding parking strategy based on the generative adversarial imitation learning through the original image data of a parking track, and enabling the parking track generated by the generated parking strategy in an actual parking process to be similar to a successful parking track. Due to the fact that the method is an online learning algorithm, multiple experiments can be carried out while learning is carried out, multiple failures can be experienced before an excellent parking strategy is successfully learned, but the failed parking track data can be stored for further learning, and therefore the learning speed can be increased, and the sample utilization rate can be increased. The learned intelligent parking strategy is not based on rules, but is a relatively intelligent strategy, so that the intelligent parking strategy can be qualified for automatic parking in different scenes.
本发明涉及一种基于生成对抗模仿学习的自动泊车方法及系统,利用泊车轨迹的原始图像数据,基于生成对抗模仿学习生成相应的泊车策略,且生成的泊车策略在实际泊车过程中产生的泊车轨迹应该与成功的泊车轨迹相似。本申请由于是一种在线学习算法,在学习的同时会进行很多次实验,并且在成功学习到优秀的泊车策略前会经历很多次失败,但是可以将这些失败的泊车轨迹数据存储下用于进一步的学习,这样能够加快学习速度并提高样本利用率。本发明由于学得的智能泊车策略不是基于规则的,而是一种较为智能的策略,因此使其能够胜任不同场景下的自动泊车。
Automatic parking method and system based on generative adversarial imitation learning
基于生成对抗模仿学习的自动泊车方法及系统
2020-06-30
Patent
Elektronische Ressource
Chinesisch
IPC: | B60W CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION , Gemeinsame Steuerung oder Regelung von Fahrzeug-Unteraggregaten verschiedenen Typs oder verschiedener Funktion / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen |
Wiley | 2023
|