Claims:We Claim:
1. A vehicle fleet monitoring system comprising:
a. a sensor data bus connected to vehicle components,
b. vehicle and driver data collected from the sensor data bus,
c. an onboard data stream mining and management module,
d. computing patterns the vehicle defined by the following data types used for representing statistical models generated onboard:
e. Telematics module
2. The method of claim 1, wherein the Input velocity data enters the system and decides a phase transition.
3. The method of claim 2 wherein the data is sent back around and is mixed with new input velocity data landmark type is decided.
4. The method of claim 3, wherein a segment must be traversed several times, and the landmarks found during each traversal are compared.
, Description:Technical Field of the Invention
The present invention relates to multi-agent systems, distributed data stream mining, and privacy-preserving data mining for mobile and distributed mining of continuously generated vehicle data streams.
Background of the Invention
No methods currently exist for multi-agent, distributed, privacy-preserving data stream mining system for characterizing vehicle, driver, and fleet monitoring. Existing monitoring systems work by downloading the data over wireless networks and then applying relatively simple linear threshold-based techniques for detecting unusual patterns.
Using data mining techniques for vehicle condition monitoring is a known art. Onboard driver performance measurement by mounting several sensors is also known. Such known systems, however, are directed primarily to performing vehicle diagnostics, assessing vehicle performance, or using sensors onboard to store the data on onboard systems and connecting the vehicle onboard computer to a remote computer for transmitting the data and visualizing it. There is no software that runs onboard a vehicle on a PDA or an embedded device and that uses lightweight data stream management and mining techniques for detecting driver's signature and continuously monitors as does the subject vehicle driver signature detection system.
A microprocessor module detachably coupled to the vehicle mounting unit affixed to and uniquely designated for a given host vehicle poles each vehicle sensor of that host vehicle to read, process, and store the vehicle operation data generated thereby. A playback mounting unit at a remote computer connects the remote computer to the host vehicle's microprocessor module in order to establish digital communication whereby the vehicle operation data and the analysis results processed therein are retrieved and displayed for a user. In addition, the driver integrity-checking module is based on some pre-determined values of the parameters and is done remotely after the data is played back on the remote computer. Also, the vehicle needs to be mounted by a multiple number of sensors as opposed to using the standard OBDII data bus for getting the vehicle data in the subject vehicle driver performance system.
Various transducers for continually monitoring various vehicle parameters are employed in that system; however, comprehensive means for analyzing the measured vehicle parameters to characterize or assess driver performance, per se, are not provided.
Prior state-of-the-art is based on linear threshold-based techniques that allow simple tasks such as detection of a feature value crossing a limit set a priori. Moreover, these techniques are applied after the data is uploaded to a remote desktop computer from the vehicle. For example, these techniques may check whether the driver crossed a specified speed limit. Unfortunately, these techniques are not capable of detecting linear and nonlinear complex driving patterns and they require an expensive process of transferring data to a remote monitoring station at a regular basis over the wireless network.
Needs exist for improved systems using mobile and distributed data stream management and mining algorithms for mining continuously generated data from different components of a vehicle.
Object of the Invention
The present invention is a method and system using mobile and distributed data stream mining algorithms for mining continuously generated data from different components of a vehicle.
Summary of the Invention
The system is designed for both onboard or remote mining and management of the data in order to characterize the typical and unusual vehicle health, driving, and fleet behavior. The system uses resource-constrained lightweight data stream management, stream mining, distributed data mining, and privacy-preserving data mining techniques.
The present approach is based on advanced multi-variate data stream mining techniques that work using the following collection of general technology categories: i) Data stream mining, ii) Distributed data stream mining, iii) Privacy-preserving data mining. The current approach offers major improvement in capabilities on two grounds. First, avoiding the expensive process of uploading the data generated by the vehicle continuously to a remote desktop-based monitoring computer over a wireless network. The approach dramatically cuts down the operating cost of such a driver characterization and monitoring system. Second, it offers advanced analytical capabilities for driver characterization and monitoring that work onboard the vehicle.
The current invention is a multi-agent distributed vehicle data mining software system that executes data stream mining methods for extracting the underlying patterns hidden in the continuous stream of data generated by the different vehicle components in multiple vehicles connected over a data communication network.
The system has four main components. The first component is an interface for the on-board diagnostic (OBD-II) data bus that couples with the software system. The system may also be connected with onboard GPS module and other sensors.
Brief Description of Drawings
FIG. 1 is a data flow of landmark identification algorithm.
Detailed Description of Invention
This algorithm is outlined in FIG. 1. Input velocity data enters the system and decides a phase transition. If not, the data is sent back around and is mixed with new input velocity data. If yes, a landmark type is decided. Results are sent back to the loop or are sent to a decision module. The decision module evaluates data near phase transition to classify the landmark type. Results are then sent back to the landmark type decision.

If a segment is traversed several times from point A to B, the velocity data is not expected to be the same every time. Random events such as the color of a traffic light, road congestion, surrounding drivers, etc. affects the vehicle's behavior. In order to accurately define the landmarks on a segment a segment must be traversed several times, and the landmarks found during each traversal are compared. The comparison allows differentiation between a stop sign and a traffic light or between a slow-moving side street and a heavily congested highway.
While the invention has been described with reference to specific embodiments, modifications and variations of the invention may be constructed without departing from the scope of the invention.