Interval Split Covariance Intersection Filter: Theory and Its Application to Cooperative Localization in a...

Area of Science:

• Robotics and Autonomous Systems
• Sensor Fusion and Estimation Theory

Background:

• Data incest, characterized by inter-estimate correlation, leads to inconsistent results in data fusion.
• This problem is particularly severe in multi-sensor multi-vehicle (MSMV) systems, causing pessimistic estimations and increased computational load.

Purpose of the Study:

• To propose a novel data fusion method, the interval split covariance intersection filter (ISCIF), to mitigate the data incest problem.
• To design a decentralized MSMV localization system incorporating both absolute and relative positioning stages.

Main Methods:

• Development and theoretical consistency proof of the interval split covariance intersection filter (ISCIF).
• Implementation of a decentralized MSMV localization system with absolute positioning using ISCIF and relative positioning.
• Utilization of interval constraint propagation (ICP) for preprocessing relative position estimates before ISCIF application.

Main Results:

• The proposed ISCIF algorithm demonstrates proven general consistency.
• Comparative simulations show that the ISCIF method achieves accurate and consistent localization results in decentralized MSMV systems.
• The method effectively addresses the data incest problem, outperforming state-of-the-art approaches.

Conclusions:

• The interval split covariance intersection filter (ISCIF) is an effective solution for the data incest problem in MSMV systems.
• The proposed decentralized localization system enhances both accuracy and consistency in vehicle positioning.
• ISCIF offers a significant advancement in sensor fusion for autonomous vehicle applications.