Unsupervised anomaly detection for autonomous vehicles

• Main Authors: CHANDWANI, VIJAY, JONES BEVAN L, SIDDAMEID, HERBERT, KOROMANSKI KRZYSZTOF
• Format: Patent
• Language: Chinese; English
• Published: 15.04.2022
• Subjects: AEROPLANES; AIRCRAFT; AVIATION; CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; CONTROLLING; COSMONAUTICS; COUNTING; DESIGNING, MANUFACTURING, ASSEMBLING, CLEANING, MAINTAINING ORREPAIRING AIRCRAFT, NOT OTHERWISE PROVIDED FOR; ELECTRIC DIGITAL DATA PROCESSING; GROUND OR AIRCRAFT-CARRIER-DECK INSTALLATIONS SPECIALLYADAPTED FOR USE IN CONNECTION WITH AIRCRAFT; HANDLING, TRANSPORTING, TESTING OR INSPECTING AIRCRAFTCOMPONENTS, NOT OTHERWISE PROVIDED FOR; HELICOPTERS; PERFORMING OPERATIONS; PHYSICS; REGULATING; SIGNALLING; SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES; TRAFFIC CONTROL SYSTEMS; TRANSPORTING

In some embodiments, techniques are provided for analyzing time series data to detect anomalies. In some embodiments, a machine learning model is used to process time series data. In some embodiments, a machine learning model is trained on a large amount of previous time series data in an unsupervised manner, allowing for the creation of a high precision model from new data. In some embodiments, the training of the machine learning model alternates between fitting optimization and pruning optimization to allow processing of a large amount of training data including unmarked anomaly records. Because machine learning models are used, anomalies within complex systems, including but not limited to autonomous vehicles such as unmanned aerial vehicles, can be detected. When an anomaly is detected, a command can be sent to a monitored system, such as an autonomous vehicle, to deal with the anomaly. 在一些实施例中，提供了用于分析时间序列数据以检测异常的技术。在一些实施例中，使用机器学习模型来处理时间序列数据。在一些实施例中，机器学习模型以无监督方式在大量先前时间序列数据上训练，从而允许从新数据中创建高精确模型。在一些实施例中，机器学