Autonomous Vehicle Data Recorder Power Loss Resilience
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in preserving sensor data during accidents due to power loss, which can result in data loss and hinder analysis for accident review and ADAS improvement, as existing non-volatile memories have limited endurance and retention periods.
Innovation Solution
A data recorder system with separate cyclic buffers for accident and training sensor data, utilizing high-endurance non-volatile memory like cross-point memory and a backup power source to ensure data preservation and transmission, even during power disruptions, using a combination of volatile and non-volatile memory for efficient data buffering and retention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If non-volatile memory is used to preserve sensor data during power loss, then data retention reliability is improved, but memory endurance is limited due to finite program-erase cycles
Solution Approach 1:
The memory system is segmented into multiple storage areas with different retention characteristics. Volatile memory (DRAM/SRAM) is used for high-speed buffering during normal operation, while non-volatile memory (flash/EEPROM) provides backup storage. This segmentation allows the system to leverage the speed of volatile memory while using non-volatile memory only when needed, thereby extending overall system reliability without exhausting non-volatile memory endurance.
Solution Approach 2:
A memory management controller acts as an intermediary between sensors and storage media. It intelligently directs data writing operations, prioritizing volatile memory for frequent updates and transferring critical data to non-volatile memory only when necessary (e.g., during detected anomalies or power loss events). This mediation reduces unnecessary program-erase cycles on non-volatile memory, extending its service life while maintaining data retention reliability.
2Speed
If volatile memory is used for high-speed data buffering, then data buffering speed is improved, but data loss occurs during power loss
Solution Approach 1:
The system performs preliminary actions by continuously monitoring data integrity and power status. Before power loss occurs, the memory management controller detects anomalies (such as sudden power drops or system failures) and proactively transfers critical buffered data from volatile memory to non-volatile memory. This preliminary action ensures that important sensor data is preserved before the volatile memory loses its contents, maintaining both high buffering speed during normal operation and data preservation reliability during power loss.
3Productivity
If separate cyclic buffers are implemented for accident and training data, then data organization and retrieval efficiency is improved, but device complexity increases
Solution Approach 1:
The memory management controller implements a universal buffer management system that handles both accident data and training data through a single, unified interface. The controller automatically identifies data types based on sensor inputs and system state, routing appropriate data to designated buffer areas without requiring separate complex management logic for each data type. This multi-functional approach improves retrieval efficiency while minimizing the increase in device complexity.
Data Source
AI summary
Systems, methods and apparatus to collect sensor data generated in an autonomous vehicle. Sensors in the vehicle generate a sensor data stream during operations of the vehicle on a road. An advanced driver assistance system (ADAS) of the vehicle uses the sensor data stream to operate the vehicle and generate a trigger signal in response to a fault in object detection, recognition, identification or classification and/or in response to the detection/prediction of an accident. A cyclic buffer buffers at least a portion of the sensor data stream. In response to the trigger signal, a selected segment of the sensor data stream is stored into a non-volatile memory. The selected segment can be partially before the trigger signal and partially after the trigger signal; and selected segment can be longer than what can be fully buffered in the cyclic buffer at the time of the trigger signal.