Off-Board Sensor Sequence Prediction for Road Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional change detection techniques for vehicle navigation face challenges such as incorrect filtering of objects of interest, difficulty in identifying rare or uncommon changes, limitations due to spatial resolution of onboard sensors, and resource-intensive training requiring human annotation, which can lead to inadequate detection of small objects and inefficient navigation.
Innovation Solution
An infrastructure system with static infrastructure pods along the road, equipped with sensors and computation systems, uses a sequence to sequence neural network to generate predicted sensor data and compare it with ground truth data to detect anomalies, providing navigation guidance to vehicles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional change detection techniques are used with onboard sensors, then the system can detect environmental changes, but the spatial resolution is insufficient and small objects are missed
Solution Approach 1:
The patent introduces an offboard infrastructure system with high-resolution sensors as an intermediary between the environment and the vehicle's navigation system. This external system captures detailed environmental data and processes it to generate change detection information, which is then provided to the vehicle. This mediator approach allows the vehicle to benefit from superior detection capabilities without carrying equally sophisticated sensors onboard.
2Adaptability or versatility
If conventional object detection methods are used, then common objects can be identified, but rare or uncommon changes are difficult to detect
Solution Approach 1:
The system performs preliminary capture of comprehensive environmental data by the offboard infrastructure before the vehicle arrives. The infrastructure system continuously monitors and records environmental states, creating a baseline database of normal conditions. When the vehicle approaches, the system can quickly compare current data against this pre-established baseline to identify rare or uncommon changes, improving both detection coverage and reliability for anomalous events.
3Measurement precision
If sequence to sequence neural network is used for anomaly detection, then detection accuracy is improved, but computational resources and training complexity increase
Solution Approach 1:
The patent divides the complex anomaly detection task into separate functional segments: the offboard infrastructure system handles data capture, sequence to sequence neural network processing, and change detection; the vehicle's navigation system handles receiving and acting on the detection results. This segmentation allows the computationally intensive AI processing to occur offboard where resources are abundant, while the vehicle maintains a simpler, more reliable system.
4Speed
If onboard sensors are used for environment perception, then real-time detection is possible, but detection of small objects and rare events is inadequate
Solution Approach 1:
The patent merges the real-time detection capability of onboard sensors with the high-precision detection capability of offboard infrastructure sensors. The system combines data from both sources, using the infrastructure system's superior resolution to detect small objects and rare events that onboard sensors might miss, while maintaining real-time responsiveness through coordinated operation. This merging creates a synergistic system that achieves both speed and precision.
Data Source
AI summary
Methods, systems, and non-transitory computer readable media are configured to perform operations comprising capturing a first sequence of captured data associated with a first time window; generating a second sequence of generated data associated with a second time window based on the first sequence of data; identifying a difference between the second sequence of generated data and a ground truth sequence of captured data associated with the second time window; determining whether the difference between the second sequence of generated data and the ground truth sequence of captured data satisfies a selected threshold value; detecting a change associated with an environment when the difference between the second sequence of generated data and the ground truth sequence of captured data satisfies the selected threshold value; and based on the detected change, providing navigation guidance to a vehicle travelling in the environment.


